CA1164094A - Non-contact measurement of surface profile - Google Patents
Non-contact measurement of surface profileInfo
- Publication number
- CA1164094A CA1164094A CA000379180A CA379180A CA1164094A CA 1164094 A CA1164094 A CA 1164094A CA 000379180 A CA000379180 A CA 000379180A CA 379180 A CA379180 A CA 379180A CA 1164094 A CA1164094 A CA 1164094A
- Authority
- CA
- Canada
- Prior art keywords
- normalized signal
- wavelength
- color pattern
- sensor signals
- wavelength bands
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Expired
Links
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01B—MEASURING LENGTH, THICKNESS OR SIMILAR LINEAR DIMENSIONS; MEASURING ANGLES; MEASURING AREAS; MEASURING IRREGULARITIES OF SURFACES OR CONTOURS
- G01B11/00—Measuring arrangements characterised by the use of optical techniques
- G01B11/24—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures
- G01B11/25—Measuring arrangements characterised by the use of optical techniques for measuring contours or curvatures by projecting a pattern, e.g. one or more lines, moiré fringes on the object
- G01B11/2509—Color coding
Abstract
RD-123??
NON-CONTACT MEASUREMENT OF SURFACE PROFILE
Abstract of the Disclosure A parallax method of wavelength labeling is based on optical triangulation. A complementary color pattern projected onto the surface is characterized by a continuous variation of the power ratio of two wavelength bands, and the profile can be measured at all points in the field of view. Shifts of the wavelength bands on separate detector arrays correspond to profile deviations. A signal processor calculates a normalized signal that is independent of surface reflectivity and roughness variations; the phase shift of this signal yields depth data from which the surface profile can be mapped.
NON-CONTACT MEASUREMENT OF SURFACE PROFILE
Abstract of the Disclosure A parallax method of wavelength labeling is based on optical triangulation. A complementary color pattern projected onto the surface is characterized by a continuous variation of the power ratio of two wavelength bands, and the profile can be measured at all points in the field of view. Shifts of the wavelength bands on separate detector arrays correspond to profile deviations. A signal processor calculates a normalized signal that is independent of surface reflectivity and roughness variations; the phase shift of this signal yields depth data from which the surface profile can be mapped.
Description
1~64/294 NON-CONTACT M~ASU~MENT 0~ SUI~FACE PRO~ILE
Background of the Invention This invention relates to methods and apparatus for measuring surface profile by the use of color and with relative immunity to surface reflectivity and roughness.
It has been known for many years that optical triangulation can yield accurate knowledge of surface profile. The basic idea is shown in FIG. 1. The shift in observed position, ~, of the incident pencil ray in intersection with the surface, allows the calculation of the shift in position of the surface with respect to the reference surface (z=O). That is, ~ = ~zM Sin a where M is the optical magnification, 9 is the parallax angle, ~z is the shift in surface position, and ~ is the shift in observed spot position. This concept has been employed in a wide variety of metrological sensors, but has a number of drawbacks. The measure-ment involves only one point at a time and the maximum rate is 10,000 points/sec. Accurate scanning of the incident ray must be maintained. A higher scan rate is possible if there are several incident rays, but then there is ambiguity as to which relected ray is which on the sensor array. The result is dependent on surface reflectivity which can vary several orders of magnitude with metal surfaces.
In the application of surface topography the use of parallax photography (photogrammetry) has overcome the first of these objections.
In this case each point in one of thc two vie~ls of a surface is ~1~4'~
correldted with the correspondiny neighborhood in the othcr view.
The location of the correlation peak allows the calculation of the shiil between the two views, which in turn leads to surface profile. This approach has the drawback that the correlation is computationally expensive and leads to some ambiguity if there is uncorrelated noise present in the images.
A final relevant approach is the so-called Moire fringe technique. A bar pattern is projected onto the surface to be measured. The surface is observed through a similar bar mask. This leads to interference fringes that indicate the vàriation in surface profile. The drawback is that the lateral resolution cannot exceed the spacing of the bar pattern. In addition, the variation in surface reflectivity inteferes with fringe formation.
Summary of the Invention The parallax method with wavelength labeling involves project-ing onto the surface of a reflective object a color pattern having at least two wavelengths or wavelength bands of light, in the u~ltraviolet, visible, or infrared regions. This pattern provides a well defined spatial variation so that shifts of reflected light due to profile variations can be easily detected. The use of different wave-lengths provides immunity to surface reflectivity variations in that shifts correspond to profile deviations. It is preferred to have a complementary color pattern which is fonned by extracting equal energy bands from a light source. The position of each point in the projected pattern is uniquely identified by the relative amounts of each wavelength power incident on the surface. In order to have depth sensitivity the power variation should be steep; periodic saw-tooth or cosine light transmissions satisfy the requirement.
. 1164'~294 The use of wavelen-Jth labelin~J mdkes the identification of shifts straiyht forwdrd. Reflected ~ight is sensed at a given parallax angle and the different wavelengths or wavelength bands are separately detected and shift position on the detector array depending upon thè depth from the object surface to a reference.
Separate sensor signals are generated corresponding to the detected light intensity of each wavelength band. The sensor signals are combined in the color pattern decoupling circuitry of a signal processor and a normalized signal is extracted that is substantially independent of surface reflectivity and roughness variations. The phase shift of the normalized signal with respect to a reference is determined and hence depth data from which the surface profile can be mapped.
The optical sensor given in detail has a transmitter with a multispectral lamp, dichroic mirrors for extracting visible and infrared bands, condensing lenses, and a patterned mirror with a bar pattern that combines the wavelength bands and projects a periodic complementary color pattern. The optical receiver has dichroic filters for extracting the visible and IR bands from the reflected light and separate linear photodiode detector arrays. The normalized signal is calculated by the equation (Ya'lVa-Vb'/Vb), where Va' and Yb' are the first derivatives of sensor signals Ya and Vb. A peak and zero detector and an interpolator determines the phase shift periodically, and depth data is produced corresponding to the phase shift.
Brief Description of the Drawinqs FIG. 1 is a prior art diagranl of the basic concept of single point optical triangulation;
6~n94 R~-12~!Q
~IG. 2 depicts the principle of the pdrallax n~thod with wave length labeling;
FIG. 3 shows a linear ramp relationship between P~1/total power and position;
FIG. 4 shows the normalized linear ramp with a deviation due .
to a profile step;
FIG. S is a plot of complementary "sawtooth" light transmissions for wavelengths A1 and A2;
FIG. 6 is a simplified diagram of the preferred embodiment of the optical sensor for generating and projecting a complementary color pattern onto an airfoil part, and for separately detecting the received color bands;
FIG. 7 is a graph of the energy distribution of a tungsten light source;
FIG. 8 is a plan view of the patterned mirror in FIG. 6;
FIG. 9 illustrates the two linear photodiode arrays with relation to the condensing lenses and patterned mirror;
FIG. 10 is a waveform diagram of complementary "cosinusoidal"
~isible and IR light transmissions;
FIGS. 11 and 12 show the shifted visible and IR received light waveforms;
FIG. 13 is a system block diagram of the non-contact profile sensor ~.the flaw detector is not part of this invention);
FIGS. 14a and 14b are detailed block diagrams of the preferred enlbodiment of the color pattern decoupling circuitry and signal shape to profile mapping circuitry;
FIG. 15 is a diagram for explaining the color pattern decoupling circuitry;
FIG. 16 is a wavefornl diagram of the normalized signal illustrdt-ing the phase shift from the expected sisnal due to a change in surface profile;
1~6~ 34 FIG. 1~ is a diaqram for explaining the pedk detector circuit, and FIG. 18 is used in profile mapping to determine z, the depth or profile data, given x, the phase shift data.
Description of the Preferred Embodiments The main difficulty in any parallax triangulating method is the location of matching points in the projected and sensed light patterns. With a single point, the problem is eliminated at the expense of speed. The wavelength labeling technique illustrated in FIG. 2 provides many points in parallel without ambiguity. Here two incident rays are provided, L and L'. The rays are made distinct by providing different colors, say blue and red, and reflected rays R and R' are focused by objective lens 20 onto a detector array 21. The spot images are now distinct if the detector consists of two arrays with color separating filters. Another benefit is readily obtained, insensitivity to reflectivity variations. Color -is the parameter of interest and this can be derived from the two detector signals independent of absolute signal level. The only remaining unwanted sensitivity is to color variations of the surface being measured. This is not of great concern with many objects, such as airfoil parts, which generally provide uniform wavelength reflectance properties. In any case, the color variations are less fre~uent and more predictable than reflectivity variations.
The concept is valid for two or more colors or for a continuous variation of wavelengths.
The next step is to consider a simple spatial variation in wavelength. Suppose that two discrete wavelengths, A1 and A2, are provided in tlle illumillation source and that, as shown in ~64~94 RD-12319 FIG. 3, these are then combined. The sum of the two wavelength powers is assumed to be constant, and the ratio P~l/total power or P~2/total power is continuously variable and is linear. The posi-tion of each point in the light pattern is uniquely identified by the relative amounts of each wavelength power incident on the surface. Assuming that the detector arrays are sensitive to only one of the wavelengths, it is possible to compute a ratio of the detected light signals and get a unique color position. The ratio of the difference in their signals to the sum is related directly to the spatial variation of the above ratio. The simplest way to show this is as follows:
P~l = kx/W
P~2 = k(l-x/W) P~l + P~2 = k p~1 + p~2 = 2x/W-l where k = a constant, x = position, W = maximum width.
The expression in the final equation above refers to received light power and is known as the normalized signal. The normalized signal, Vn, is plotted in FIG. 4 and would be a straight line if the sur-face whose profile is being measured were flat. A step change in the surface delays the waveform by a certain amount.
Thus, the deviation and position on the sensor of a particular ratio of wavelength powers leads directly to shift in surface position. Since the illumination produces a continuous variation ofpower ratios,the profile can be measured at each point in the sensor. An important feature is that it can be shown by a similar mathematical treatment that the normalized signal is independent of surface reflectivity so that variations in this reflectivity and in surface roughness do not affect the resulting shift measurement. This assumes that the reflectivity is inde-pendent of wavelength. The reflectivity appears in both the 116~)5`4 difference and sum tenns and cancels out when the ratio is taken.
The wavelength labeling technique using a single linear ramp variation of wavelength powers has two constraints that cannot both ~e satisfied. A linear variation is wanted and also a high sensitivity to profile changes, but these two are incompatible; A high sensitivity to profile changes calls for a high slope but then the maximum width, W, that can be measured at one time becomes small. The sawtooth or equilateral triangle light transmission pattern in FIG. 5 has good depth sensitivity and a larger field of view can be covered at one timt-~
This is a periodic, complementary color pattern;~at every point, x,the total transmitted power, P~1 + P~2, is constant. There is a limit as to how close the periods can be, and the period cannot become too short because the problem of ambiguity arises again. In the received light pattern, the peaks shift in response to profile changes, and the peaks begin to cross if the period is too short. On the other hand, the period should not be too large because a steep slope is wanted. There is a compromise, then, as to ambiguity and depth sensitivity, and the slope steepness of the sawtooth transmitted light pattern is selected with this in mind. The sawtooth complementary color pattern can be imptemented but may require well collimated light beams such as are produced by laser light sources.
The preferred embodiment of the optical sensor is depicted in FIG. 6. This sensor was developed to achieve a number of specific performance goals in accordance with the requirements for inspection of metal surfaces, in particular the profile measurement accuracies involved in aircraft engine airfoil parts, or turbine blades. The technique is applicable to many reflecting surfaces and the sensor represents a general solution to the problem of profile mcasurellent. Having dctermincd thc surface profile, it is then possible to look for local deviations in the surface such as pits, - 1164~94 K~ lu fractures, d~?nts, nic~:s, and disclordtions. The performance goals are that it is desired to detect 0.001 inch variations in surface profile over a one inch field of view. Surface defects with lateral dimensions as small as 0.010 inch are to be detected. The reflectivi~y of part surfaces may vary as much as an order of magnitude due to coatings and surface finish. Sensor data rates of 106 pixels/sec are to be provided. This optical sensor which implements the wave-length labeling technique meets all of these specifications.
In FIG. 6, the sensor is comprised of an optical transmitter 22 and an optical receiver 23. The parallax angle is chosen - carefully and 26 is a good compromise; at this angle there is adequate sensitivity to profile changes and reduced sensitivity to local surface roughness variations. Light source 24 is an incandescent tungsten lamp, which is a multispectral source with illumination in the visible and infrared regions. The infrared portion of the illum.nation (~ > 8000 A) represents about 50 percent of the available power, and to achieve maximum illumination efficiency, this energy must be used. Elliptical reflector 25 is selected to reflect all wavelengths emitted by the source. The spectral distribution of the tungsten lamp can be conveniently split into two equal wavelength bands by readily available dichroic reflectors, and it is desirable to have equal powers in both wavelength bands. Referring to FIG. 7, which shows the tungsten source distribution, there are many possible choices, one in the visible region and the other in the infrared region, satisfying the requirements that the bands can be separated by available optical filtering techniques and that the powers are equal.
Another rcquiremcnt is that the source and light detector should be well matched. The power distribution from the tungsten lamp is well matched to the solid state detector.
~ 4~94 i~D-1~31~!
~ dichroic mirror 26, I-IG. ~, reflects the IR spectral colnponents dnd translnits the visible components. A second dichroic mirror 27 has a light absorbing backing 28 and reflects only the extracted IR wavelength band to an aspheric condensing lens 23.
The visible light is reflected by another dichroic mirror 30 having a light absorbing backing 31, and the extracted visible wavelength band passes through an aspheric condensing lens 32. A patterned mirror 33 has a bar pattern 34 on one surface which is depicted in greater detail in FIG. 8. Bars 34 have equal widths and are equally spaced from one another, and can be metallized stripes. The IR
band is transmitted through the spaces between the bars, FIG. 6, and the visible band is reflected by the bars. The IR and visible bands alternate and the light is registered and projected in the same direction. Objective lens 35 focuses the complementary color pattern 36 onto the surface of airfoil part 37.
The transmitted complementary color pattern is shown in FIG. 10, where it is observed that the waveforms of the visible and IR bands are "cosinusoidal", have a constant amplitude, and are 180 out of phase. It might be expected that bar pattern 34 would produce a square wave projected light pattern with alternating visible and IR stripes, but the cosine wave distribution of the two bands is actually produced because of the spreading out effect of aspheric lenses 29 and 32. The projector optics produces a circular illumination pattern but the usable part of the pattern is a central rectangle 1 inch long by 0.125 inch wide. Each wave-length power in the projected color pattern, FIG. 10, continuously varies and the total wavelength power is approximately constant at every point. As wdS the case with sawtdoth illumination, the position of eacll point in the projected pattern is uniquely identified by the relative anlo(lnts of each waveleng~h power incident 1~6~ 94 Rn-l?~
on the surface. The complementdry color pattern does more than suppl~
illumination because additionally a signal is transmitted by the projection optics.
Optical receiver 23 has an objective lens 38 and an IR
reflecting dichroic mirror 3 which transmits the visible ~avelength bands of the reflected light pattern. These bands are passed by a visible pass filter 40 and are sensed by a linear photodiode array 41 whose individual elements generate sensor signals corresponding to the detected light intensity. The IR wavelength bands are reflected to an IR pass filter 42 and are sensed by a second linear photodiode array 43 which is orthogonal to and aligned with the first array 41. FIG. 9 shows the separate detector arrays 41 and 43 and (see FIG. 8) their relation to patterned mirror 33 and condensing lens 29 and 32. The detector arrays both have 512 diodes to realize the required field of view, and are commercially available components such as the linear diode array scanner of Reticon Corp., which is electrically scanned at a l MHz clock rate. The array elements are sampled sequentially in the nature of a raster scan and the individual photodiodes generate an analog signal proportional to the light falling on that element.
Separate visible and IR sensor video signals are generated.
As the scan is performed, airfoil part 3i and optical sensor 22, 23 move relative to one another to scan all of the part surfaces. One end of the part is clamped mechanically and then the other end so that profile data is ta~en all over the part. The color illumination pattern is incident approximately normal to the surface at all times, and optical transmitter and receiver 22 and 23 are fixed relative to one another and the parallax angle is unchallyed. Charlges in depth of the surface cause a corresponding ci~dnge in phase be~een the trans;nitted ~6~9~ ~
RD-1~319 and rec~ive~ light p~t~erns. In the ~ptical receiver~ the visible and IR wavelength bands shift position on detector arrays 41 and 4~ depending upon the depth from the part surface to a reference surface, which reference surface is in the optical sensor. FIGS. 11 S and 12 illustrate shifts in the peaks of the visib~e and IR detected light or received power waveforms due to profile variations. The shift is the same in both waveforms at any given position. The double ended arrows indicate that the shifts of the peaks at other x positions depends upon the profile variation at those points.
Before proceeding further7 alternative approaches for realizing the complementary color pattern will be mentioned. Laser and arc light sources are also suitable. Lasers produce nearly single wave-length illumination and two different lasers are required. The argon ion and neodymiun yag types have the required power level, but require arc tube excitation and attendant complex power supplies. The problem of spec'~l~ can be overcome, but the sources are too cumbersome physically to be mounted directly in the sensor head assembly. The most readily available arc source for the present application is the indium-argon system. This lamp has a suitable wavelength distribution and more than adequate illumination power, but has an unsatisfactory life and needs to be cooled. Instead of patterned mirror 33, film spatial patterns may be used. Two such patterns are prepared and the transmissions of the film patterns are complementary. An output beam splitter recombines the wavelength bands, and a disadvantage is the requirement to align the two film patterns and the beam splitter. Another disadvantage is the film itself; the proper amplitude variation depends on film exposure and development conditions and these are difficult to control.
Mirror patterns are prepared by vacuum depositing an aluminum film on d slass substrate and thcn selectively etchins the film R~
usin~ photolitoJrd~)hic techni(lues. If a discrete lin~ pdttern is on a scale sl~lallcr than the resolution of the source-receiver optics, the function will be related to the spatial average of line spacing and width. This can be easily varied in a continuous fashion to realize a sawtooth transmission pattern. A diamond pattern mirror is based on the use of cylindrical lens elements. The basic idea is to transmit and reflect well-collimated beams from the mirror pattern~ The emerging spectral mixture is brought to focus at the part surface by a cylindrical lens. The diamond pattern provides a linear sawtooth variation when all rays in a line orthogonal to the cylinder axis are brought together. The scheme requires that the rays incident on the mirror pattern be well collimated, and this suggests that the diamond pattern is most applicable to the laser source.
FIG. 13 is a simplified block diagram of the non-contact profile sensor. Visible and IR band sensor signals Ya and Yb are fed to a signal processor which generates depth or profile data in real time and at a rate of 106 points/sec. Color pattern decoupling circuitry 45 combines the pair of sensor signals Va and Vb, decouples color pattern data from surface function data, and extracts a single normalized signal that is substantially independent of surface reflectivity and roughness variations. The normalized signal is fed to signal shape to profile mapping circuitry 46; the - phase shift of the peaks and zeroes of the normalized signal with respect to a reference is determined, and the phase shift information is converted to depth or profile data from which the surface profile can be mapped. Flaw detector 47, which is not a part of this invention, recognizes and identifies local deviations in the surface profile, such as pits, fractures, dents, etc., which are a basis for rejection of the part.
R~
- The theory underlying color pattern decoupliny is set forth before discussing the preferred hardware implementation in FIG. 14a~
Va = A
~'~ ' Vb = ~ Ç313 where A and B are models of the light scattering from the surface, ancl a and ~ are models of the labeling scheme. These terms can be expressed locally as polynomials:
A = Ao + Alx 2 = aO + lX + 2X
B = Bo + B1x B = ~0 + ~lx + B2x2 Multiplying polynomi~ls, a O O (A1aO + AO1)x + (A11 + Ao2)x2 + (A12)x3 Vb Bo~o + (Bl~o, + sO~l)X + (Bl~l + B0~2)X + (B1~2)X
The foregoing are Taylor series and it can be shown that:
Va ' Al ,aO Aol Al 1 +
- Va Aoo Ao O
Vb'Bl~o + Bo~l Bl ~1 -- = +
Yb Bo~o Bo ~0 wilere Va' and Vb' are first derivatives about the relative point x=0. Then, Vd l Vb / A1 B1 / 1 ~1 \
= . . I +
a b ~ O Bo / ~oO ~0 The first bracketed term goes to zero because the surface is the same and the surface functions are identical. Thè extracted color labeling data remains:
Va' Vb 1 ~1 Va Vb ~0 ~0 This technique is more ~eneral than that given previously, calculatin~
the ratio of the difference of the sensor signais to the sum of these signals, because the latter assumes equal wavelength band powers.
The above is ~ood for equal or unequal wavelength band powers.
v ~
Referring to FIG. 14a, ~n~s~b~e and IR band detector video signals Va and Yb, in digital form, are fed to a detector array noise removal circuit 48 and hence to a memory bus 49. Both detector arrays 41 and 43 are scanned electrically at a high rate and the serialized data on both lines is continuously fed to color pattern decoupling circuitry 45. The data for five points or pixels is read out in parallel to four operators 50-53. All four operator circuits are identical and are implemented with a large scale integrated multiplier-accumulator chip (such as TRW Type TDC1009J) and a memory to store a set of coefficients for each of the five points.
Any given section of the cosine wave is approximated by a second order polynomial. The operators calculate the polynomial which is least squares fitted to a number of points about a given pixel.
Points 1-S in FI6. 15 are typical of the data that is sensed.
The problem to determine the cosine wave 54 that is the best least squares fit, i.e., the sum of the squares of the differences between the curve and each data point is a mininlum. In each 4~34 ~n operator 50-53, the stored coefficients are sequentially multiplied by the sensed pixel data and the sum for all five points is calculated With reference to the normalization equation (Va'/Va ^ Vb'/Vb), operators IX and rX calculate the derivatives of the polynomial at any given point, and operators I8 and IB calculate the values of the polyncmial at any given point. Circuits 55 and 56 perform the indicated divisions and subtractor circuit 57 generates the normalized signal ~n. These operators are applied at each pixel position sucessively in a serial raster scan fashion. The resulting values of TB ~ IB ~ Ix , and Ix are the desired function (value and derivative) of the least squares polynomial fit (of preselected order) about the pixel of evaluation.
Assuming that the surface being measured is flat, the expected normalized signal shown in dashed lines in FIG. 16 is a periodic cosine wave and the periods are equal to one another. A change in the profile, ~z, leads to a phase shift of the detected cosine wave which is shown in full lines. The phase shift at the peak of the cosine wave, and also at the zero, is proportional to the depth or profile change, ~z. The symbols P+ and P , standing for positive and negative peaks of the wave, and Z and Z , standing for positive slope and negative slope zeros of the cosine wave, are indicated in FIG. 16. The normalized signal, it is re-emphasized, is independent of surface reflectivity and roughness variations.
The mathematical basis for peak detection will be explained with reference to FIG. 17. Two oppositely sloped curve sides S1 and S2 meet at peak point xO. The following three functions tend to go to zero at the peak and their sum is a minimum at the peak, Fpeak - > 0.
F1: aa (P(xO`)) ~
At the peak the top is flat and the partial derivative of the polymonial P approximately at xO goes to zero.
11~;4~9~
,~D-;,3 F2: P(S1(xO)) - P(S2(xO)) -- ~ 0 The polynomial approximation of both sides S1 and S2 predicts the same point.
F3: Slope (P(S1)) + Slope (P(S2)) - ~
S The slope of both sides are approximately equal and opposite. The peak is located when the sum of the absolute value of these three functions, taken at many points xO, is minimum.
Turning to FIG. 14b, the normalized signal, although smoothed by the pattern decoupling process, still has localized fluctations and irregularities and may even in some cases contain false zeros and peaks. The normalized signal is fed to a memory buss 60, and sets of nine data points are read out in sequence and presented to a peak and zero detector 61. Linear operators 62-64 are identical to circuits 50-54; they respectively calculate the functions F1, F2, and F3, and their absolute sum is calculated in circuit 65 and passed to the minimum detector 66 by selector 67 which determines whether a peak or zero is being detected. This determination is made by comparing the output of the peak detector components 62-65 and zero detector components 69 and 70 and selecting the signal ~lith the smaller amplitude. This signal is then tracked until it achieves a minimum value at which point the peak or ZerQ is said to exist. When a positive or negative peak in the normalized signal is recognized, the index number of the photodioae array element (1-512) is read out of memory 68.
A set of five d3ta points is also presented to operator 69, which is identical to operators 50 and 52. This operator, I8, determines the cosine wave that is the best least squdres fit.
Absolute value circuit 70 is essentially a rectifier and sends an -lfi-Dl~
output to minilllulll(lctector 66 when a z~ro in th~ nor~ndlized signdl is identified. The index numbers corresponding to the signal zeros ire read out of memory 68. Interpolator 71 is in parallel with peak and zero detector 61, which determines the gross location of the peaks and zeros. This interpolator subdivides the phase shift pixel interval into 32 parts and gets the peaks or zeros between pixels.
The added increment ~Zero is calculated by dividing the normalized - signal, ~n' by the first derivative, Ynl; the added increment ~Peak is calculated by dividing Vn by Ynll, the second derivative. The set of nine data points is fed in parallel to circùits 72-74, which are identical to operator 69. Operator IB yields the best least squares fit cosine wave at a given point, operator IX yields the first derivative of the polynomial at the given point, and operator IXx calculates the second derivative at that point. The indicated divisions are performed by blocks 75 and 76, and the QZero or ~Peak increment is gated by a selector 77.
A classifier 78 identifies the four cosine wave classes or ~eatures P , P , Z , and Z (see FIG. 16) and the one that is selected depends on the sign of IB and Ix. The class information is fed to selector 77 and either the interpolated zero or interpolated peak data is passed to a profile mapping subsystem 79. Profile mapping circuitry 79 has a set of memories such that, given the class, a particular mcmory is accessed and the depth or profile data, z, corresponding to the phase shift can be read out. The need for four memories, one for each of the classes, Z , P , Z , and P , can be seen in FIG. 18 where it is shown that the four classes overlap. In every one of the four hprizontal sets of ramps, each of which begins where the preceding one stops, the number of ramps corresponds to the number o~ periods in the conlplelllentary color pattern. Only two are shown in order to conserve space. There is a known relationship o~
. .
~D-12319 betwe~l) phdse shift an-~ nl~asure~ ~epth, and yiven the phase shift for a particuldr cldss and period, the depth or profile data is stored in this memory. The index number (see FIG. 14b) is also transmitted to a look-up table 80, and corresponding x position data is read out~
This position is simply the location of the given label relative to a fixed frame of reference such as the label transmitter. A complete identification of the profile or depth data requires values of both x and z.
A more general method of reconstructing the surface can be implemented by defining a more refined label structure based on the shape of the original light signal. In specific, if the shape of the normalized sisnal is cosinusoidal, then the phase shift may be reconstructed at each pixel in the signal by simply determining the shift of the corresponding point in a reference signal. This process is limited by the noise in the signal normalization process which is introduced by three sources: the implementation hardware, the signal model, and the optics. The normalization scheme shown above can provide an adequate signal for reconstruction on the pixel level. Th;e peak and zero detector scheme is replaced by a suitable scaling and - 20 correlation circuit to match the normalized signal with a reference signal acquired from a flat surface (z=0). This correlation requires the matching of amplitudes as a function of index and, thus, is somewhat limited at the peak of the signals since the slope is low.
A combination of the Feak detector scheme with the correlation - 25 scheme provides enhanced response in these regions at the expense of some smoothing.
In conclusion, an improved concept for the non-contact measurement of surface profile has been disc!osed. A novel aspcct is the specific approach which allows the determination of profile at all points in the field of YieW. In dddition, these medsurelllents ~i64094 ``;
r~
are relatively imnlune to surface reflectivity and roughness. The color illwllindtion pattern may be composed of ultraviolet, visible, or infrared light.
While the invention has been particularly shown and described with reference to preferred embodiments thereof, it will be understood by those skilled in the art that the foregoing and other changes in form and details may be made therein without departing from the spirit and scope of the invention.
1 9- '
Background of the Invention This invention relates to methods and apparatus for measuring surface profile by the use of color and with relative immunity to surface reflectivity and roughness.
It has been known for many years that optical triangulation can yield accurate knowledge of surface profile. The basic idea is shown in FIG. 1. The shift in observed position, ~, of the incident pencil ray in intersection with the surface, allows the calculation of the shift in position of the surface with respect to the reference surface (z=O). That is, ~ = ~zM Sin a where M is the optical magnification, 9 is the parallax angle, ~z is the shift in surface position, and ~ is the shift in observed spot position. This concept has been employed in a wide variety of metrological sensors, but has a number of drawbacks. The measure-ment involves only one point at a time and the maximum rate is 10,000 points/sec. Accurate scanning of the incident ray must be maintained. A higher scan rate is possible if there are several incident rays, but then there is ambiguity as to which relected ray is which on the sensor array. The result is dependent on surface reflectivity which can vary several orders of magnitude with metal surfaces.
In the application of surface topography the use of parallax photography (photogrammetry) has overcome the first of these objections.
In this case each point in one of thc two vie~ls of a surface is ~1~4'~
correldted with the correspondiny neighborhood in the othcr view.
The location of the correlation peak allows the calculation of the shiil between the two views, which in turn leads to surface profile. This approach has the drawback that the correlation is computationally expensive and leads to some ambiguity if there is uncorrelated noise present in the images.
A final relevant approach is the so-called Moire fringe technique. A bar pattern is projected onto the surface to be measured. The surface is observed through a similar bar mask. This leads to interference fringes that indicate the vàriation in surface profile. The drawback is that the lateral resolution cannot exceed the spacing of the bar pattern. In addition, the variation in surface reflectivity inteferes with fringe formation.
Summary of the Invention The parallax method with wavelength labeling involves project-ing onto the surface of a reflective object a color pattern having at least two wavelengths or wavelength bands of light, in the u~ltraviolet, visible, or infrared regions. This pattern provides a well defined spatial variation so that shifts of reflected light due to profile variations can be easily detected. The use of different wave-lengths provides immunity to surface reflectivity variations in that shifts correspond to profile deviations. It is preferred to have a complementary color pattern which is fonned by extracting equal energy bands from a light source. The position of each point in the projected pattern is uniquely identified by the relative amounts of each wavelength power incident on the surface. In order to have depth sensitivity the power variation should be steep; periodic saw-tooth or cosine light transmissions satisfy the requirement.
. 1164'~294 The use of wavelen-Jth labelin~J mdkes the identification of shifts straiyht forwdrd. Reflected ~ight is sensed at a given parallax angle and the different wavelengths or wavelength bands are separately detected and shift position on the detector array depending upon thè depth from the object surface to a reference.
Separate sensor signals are generated corresponding to the detected light intensity of each wavelength band. The sensor signals are combined in the color pattern decoupling circuitry of a signal processor and a normalized signal is extracted that is substantially independent of surface reflectivity and roughness variations. The phase shift of the normalized signal with respect to a reference is determined and hence depth data from which the surface profile can be mapped.
The optical sensor given in detail has a transmitter with a multispectral lamp, dichroic mirrors for extracting visible and infrared bands, condensing lenses, and a patterned mirror with a bar pattern that combines the wavelength bands and projects a periodic complementary color pattern. The optical receiver has dichroic filters for extracting the visible and IR bands from the reflected light and separate linear photodiode detector arrays. The normalized signal is calculated by the equation (Ya'lVa-Vb'/Vb), where Va' and Yb' are the first derivatives of sensor signals Ya and Vb. A peak and zero detector and an interpolator determines the phase shift periodically, and depth data is produced corresponding to the phase shift.
Brief Description of the Drawinqs FIG. 1 is a prior art diagranl of the basic concept of single point optical triangulation;
6~n94 R~-12~!Q
~IG. 2 depicts the principle of the pdrallax n~thod with wave length labeling;
FIG. 3 shows a linear ramp relationship between P~1/total power and position;
FIG. 4 shows the normalized linear ramp with a deviation due .
to a profile step;
FIG. S is a plot of complementary "sawtooth" light transmissions for wavelengths A1 and A2;
FIG. 6 is a simplified diagram of the preferred embodiment of the optical sensor for generating and projecting a complementary color pattern onto an airfoil part, and for separately detecting the received color bands;
FIG. 7 is a graph of the energy distribution of a tungsten light source;
FIG. 8 is a plan view of the patterned mirror in FIG. 6;
FIG. 9 illustrates the two linear photodiode arrays with relation to the condensing lenses and patterned mirror;
FIG. 10 is a waveform diagram of complementary "cosinusoidal"
~isible and IR light transmissions;
FIGS. 11 and 12 show the shifted visible and IR received light waveforms;
FIG. 13 is a system block diagram of the non-contact profile sensor ~.the flaw detector is not part of this invention);
FIGS. 14a and 14b are detailed block diagrams of the preferred enlbodiment of the color pattern decoupling circuitry and signal shape to profile mapping circuitry;
FIG. 15 is a diagram for explaining the color pattern decoupling circuitry;
FIG. 16 is a wavefornl diagram of the normalized signal illustrdt-ing the phase shift from the expected sisnal due to a change in surface profile;
1~6~ 34 FIG. 1~ is a diaqram for explaining the pedk detector circuit, and FIG. 18 is used in profile mapping to determine z, the depth or profile data, given x, the phase shift data.
Description of the Preferred Embodiments The main difficulty in any parallax triangulating method is the location of matching points in the projected and sensed light patterns. With a single point, the problem is eliminated at the expense of speed. The wavelength labeling technique illustrated in FIG. 2 provides many points in parallel without ambiguity. Here two incident rays are provided, L and L'. The rays are made distinct by providing different colors, say blue and red, and reflected rays R and R' are focused by objective lens 20 onto a detector array 21. The spot images are now distinct if the detector consists of two arrays with color separating filters. Another benefit is readily obtained, insensitivity to reflectivity variations. Color -is the parameter of interest and this can be derived from the two detector signals independent of absolute signal level. The only remaining unwanted sensitivity is to color variations of the surface being measured. This is not of great concern with many objects, such as airfoil parts, which generally provide uniform wavelength reflectance properties. In any case, the color variations are less fre~uent and more predictable than reflectivity variations.
The concept is valid for two or more colors or for a continuous variation of wavelengths.
The next step is to consider a simple spatial variation in wavelength. Suppose that two discrete wavelengths, A1 and A2, are provided in tlle illumillation source and that, as shown in ~64~94 RD-12319 FIG. 3, these are then combined. The sum of the two wavelength powers is assumed to be constant, and the ratio P~l/total power or P~2/total power is continuously variable and is linear. The posi-tion of each point in the light pattern is uniquely identified by the relative amounts of each wavelength power incident on the surface. Assuming that the detector arrays are sensitive to only one of the wavelengths, it is possible to compute a ratio of the detected light signals and get a unique color position. The ratio of the difference in their signals to the sum is related directly to the spatial variation of the above ratio. The simplest way to show this is as follows:
P~l = kx/W
P~2 = k(l-x/W) P~l + P~2 = k p~1 + p~2 = 2x/W-l where k = a constant, x = position, W = maximum width.
The expression in the final equation above refers to received light power and is known as the normalized signal. The normalized signal, Vn, is plotted in FIG. 4 and would be a straight line if the sur-face whose profile is being measured were flat. A step change in the surface delays the waveform by a certain amount.
Thus, the deviation and position on the sensor of a particular ratio of wavelength powers leads directly to shift in surface position. Since the illumination produces a continuous variation ofpower ratios,the profile can be measured at each point in the sensor. An important feature is that it can be shown by a similar mathematical treatment that the normalized signal is independent of surface reflectivity so that variations in this reflectivity and in surface roughness do not affect the resulting shift measurement. This assumes that the reflectivity is inde-pendent of wavelength. The reflectivity appears in both the 116~)5`4 difference and sum tenns and cancels out when the ratio is taken.
The wavelength labeling technique using a single linear ramp variation of wavelength powers has two constraints that cannot both ~e satisfied. A linear variation is wanted and also a high sensitivity to profile changes, but these two are incompatible; A high sensitivity to profile changes calls for a high slope but then the maximum width, W, that can be measured at one time becomes small. The sawtooth or equilateral triangle light transmission pattern in FIG. 5 has good depth sensitivity and a larger field of view can be covered at one timt-~
This is a periodic, complementary color pattern;~at every point, x,the total transmitted power, P~1 + P~2, is constant. There is a limit as to how close the periods can be, and the period cannot become too short because the problem of ambiguity arises again. In the received light pattern, the peaks shift in response to profile changes, and the peaks begin to cross if the period is too short. On the other hand, the period should not be too large because a steep slope is wanted. There is a compromise, then, as to ambiguity and depth sensitivity, and the slope steepness of the sawtooth transmitted light pattern is selected with this in mind. The sawtooth complementary color pattern can be imptemented but may require well collimated light beams such as are produced by laser light sources.
The preferred embodiment of the optical sensor is depicted in FIG. 6. This sensor was developed to achieve a number of specific performance goals in accordance with the requirements for inspection of metal surfaces, in particular the profile measurement accuracies involved in aircraft engine airfoil parts, or turbine blades. The technique is applicable to many reflecting surfaces and the sensor represents a general solution to the problem of profile mcasurellent. Having dctermincd thc surface profile, it is then possible to look for local deviations in the surface such as pits, - 1164~94 K~ lu fractures, d~?nts, nic~:s, and disclordtions. The performance goals are that it is desired to detect 0.001 inch variations in surface profile over a one inch field of view. Surface defects with lateral dimensions as small as 0.010 inch are to be detected. The reflectivi~y of part surfaces may vary as much as an order of magnitude due to coatings and surface finish. Sensor data rates of 106 pixels/sec are to be provided. This optical sensor which implements the wave-length labeling technique meets all of these specifications.
In FIG. 6, the sensor is comprised of an optical transmitter 22 and an optical receiver 23. The parallax angle is chosen - carefully and 26 is a good compromise; at this angle there is adequate sensitivity to profile changes and reduced sensitivity to local surface roughness variations. Light source 24 is an incandescent tungsten lamp, which is a multispectral source with illumination in the visible and infrared regions. The infrared portion of the illum.nation (~ > 8000 A) represents about 50 percent of the available power, and to achieve maximum illumination efficiency, this energy must be used. Elliptical reflector 25 is selected to reflect all wavelengths emitted by the source. The spectral distribution of the tungsten lamp can be conveniently split into two equal wavelength bands by readily available dichroic reflectors, and it is desirable to have equal powers in both wavelength bands. Referring to FIG. 7, which shows the tungsten source distribution, there are many possible choices, one in the visible region and the other in the infrared region, satisfying the requirements that the bands can be separated by available optical filtering techniques and that the powers are equal.
Another rcquiremcnt is that the source and light detector should be well matched. The power distribution from the tungsten lamp is well matched to the solid state detector.
~ 4~94 i~D-1~31~!
~ dichroic mirror 26, I-IG. ~, reflects the IR spectral colnponents dnd translnits the visible components. A second dichroic mirror 27 has a light absorbing backing 28 and reflects only the extracted IR wavelength band to an aspheric condensing lens 23.
The visible light is reflected by another dichroic mirror 30 having a light absorbing backing 31, and the extracted visible wavelength band passes through an aspheric condensing lens 32. A patterned mirror 33 has a bar pattern 34 on one surface which is depicted in greater detail in FIG. 8. Bars 34 have equal widths and are equally spaced from one another, and can be metallized stripes. The IR
band is transmitted through the spaces between the bars, FIG. 6, and the visible band is reflected by the bars. The IR and visible bands alternate and the light is registered and projected in the same direction. Objective lens 35 focuses the complementary color pattern 36 onto the surface of airfoil part 37.
The transmitted complementary color pattern is shown in FIG. 10, where it is observed that the waveforms of the visible and IR bands are "cosinusoidal", have a constant amplitude, and are 180 out of phase. It might be expected that bar pattern 34 would produce a square wave projected light pattern with alternating visible and IR stripes, but the cosine wave distribution of the two bands is actually produced because of the spreading out effect of aspheric lenses 29 and 32. The projector optics produces a circular illumination pattern but the usable part of the pattern is a central rectangle 1 inch long by 0.125 inch wide. Each wave-length power in the projected color pattern, FIG. 10, continuously varies and the total wavelength power is approximately constant at every point. As wdS the case with sawtdoth illumination, the position of eacll point in the projected pattern is uniquely identified by the relative anlo(lnts of each waveleng~h power incident 1~6~ 94 Rn-l?~
on the surface. The complementdry color pattern does more than suppl~
illumination because additionally a signal is transmitted by the projection optics.
Optical receiver 23 has an objective lens 38 and an IR
reflecting dichroic mirror 3 which transmits the visible ~avelength bands of the reflected light pattern. These bands are passed by a visible pass filter 40 and are sensed by a linear photodiode array 41 whose individual elements generate sensor signals corresponding to the detected light intensity. The IR wavelength bands are reflected to an IR pass filter 42 and are sensed by a second linear photodiode array 43 which is orthogonal to and aligned with the first array 41. FIG. 9 shows the separate detector arrays 41 and 43 and (see FIG. 8) their relation to patterned mirror 33 and condensing lens 29 and 32. The detector arrays both have 512 diodes to realize the required field of view, and are commercially available components such as the linear diode array scanner of Reticon Corp., which is electrically scanned at a l MHz clock rate. The array elements are sampled sequentially in the nature of a raster scan and the individual photodiodes generate an analog signal proportional to the light falling on that element.
Separate visible and IR sensor video signals are generated.
As the scan is performed, airfoil part 3i and optical sensor 22, 23 move relative to one another to scan all of the part surfaces. One end of the part is clamped mechanically and then the other end so that profile data is ta~en all over the part. The color illumination pattern is incident approximately normal to the surface at all times, and optical transmitter and receiver 22 and 23 are fixed relative to one another and the parallax angle is unchallyed. Charlges in depth of the surface cause a corresponding ci~dnge in phase be~een the trans;nitted ~6~9~ ~
RD-1~319 and rec~ive~ light p~t~erns. In the ~ptical receiver~ the visible and IR wavelength bands shift position on detector arrays 41 and 4~ depending upon the depth from the part surface to a reference surface, which reference surface is in the optical sensor. FIGS. 11 S and 12 illustrate shifts in the peaks of the visib~e and IR detected light or received power waveforms due to profile variations. The shift is the same in both waveforms at any given position. The double ended arrows indicate that the shifts of the peaks at other x positions depends upon the profile variation at those points.
Before proceeding further7 alternative approaches for realizing the complementary color pattern will be mentioned. Laser and arc light sources are also suitable. Lasers produce nearly single wave-length illumination and two different lasers are required. The argon ion and neodymiun yag types have the required power level, but require arc tube excitation and attendant complex power supplies. The problem of spec'~l~ can be overcome, but the sources are too cumbersome physically to be mounted directly in the sensor head assembly. The most readily available arc source for the present application is the indium-argon system. This lamp has a suitable wavelength distribution and more than adequate illumination power, but has an unsatisfactory life and needs to be cooled. Instead of patterned mirror 33, film spatial patterns may be used. Two such patterns are prepared and the transmissions of the film patterns are complementary. An output beam splitter recombines the wavelength bands, and a disadvantage is the requirement to align the two film patterns and the beam splitter. Another disadvantage is the film itself; the proper amplitude variation depends on film exposure and development conditions and these are difficult to control.
Mirror patterns are prepared by vacuum depositing an aluminum film on d slass substrate and thcn selectively etchins the film R~
usin~ photolitoJrd~)hic techni(lues. If a discrete lin~ pdttern is on a scale sl~lallcr than the resolution of the source-receiver optics, the function will be related to the spatial average of line spacing and width. This can be easily varied in a continuous fashion to realize a sawtooth transmission pattern. A diamond pattern mirror is based on the use of cylindrical lens elements. The basic idea is to transmit and reflect well-collimated beams from the mirror pattern~ The emerging spectral mixture is brought to focus at the part surface by a cylindrical lens. The diamond pattern provides a linear sawtooth variation when all rays in a line orthogonal to the cylinder axis are brought together. The scheme requires that the rays incident on the mirror pattern be well collimated, and this suggests that the diamond pattern is most applicable to the laser source.
FIG. 13 is a simplified block diagram of the non-contact profile sensor. Visible and IR band sensor signals Ya and Yb are fed to a signal processor which generates depth or profile data in real time and at a rate of 106 points/sec. Color pattern decoupling circuitry 45 combines the pair of sensor signals Va and Vb, decouples color pattern data from surface function data, and extracts a single normalized signal that is substantially independent of surface reflectivity and roughness variations. The normalized signal is fed to signal shape to profile mapping circuitry 46; the - phase shift of the peaks and zeroes of the normalized signal with respect to a reference is determined, and the phase shift information is converted to depth or profile data from which the surface profile can be mapped. Flaw detector 47, which is not a part of this invention, recognizes and identifies local deviations in the surface profile, such as pits, fractures, dents, etc., which are a basis for rejection of the part.
R~
- The theory underlying color pattern decoupliny is set forth before discussing the preferred hardware implementation in FIG. 14a~
Va = A
~'~ ' Vb = ~ Ç313 where A and B are models of the light scattering from the surface, ancl a and ~ are models of the labeling scheme. These terms can be expressed locally as polynomials:
A = Ao + Alx 2 = aO + lX + 2X
B = Bo + B1x B = ~0 + ~lx + B2x2 Multiplying polynomi~ls, a O O (A1aO + AO1)x + (A11 + Ao2)x2 + (A12)x3 Vb Bo~o + (Bl~o, + sO~l)X + (Bl~l + B0~2)X + (B1~2)X
The foregoing are Taylor series and it can be shown that:
Va ' Al ,aO Aol Al 1 +
- Va Aoo Ao O
Vb'Bl~o + Bo~l Bl ~1 -- = +
Yb Bo~o Bo ~0 wilere Va' and Vb' are first derivatives about the relative point x=0. Then, Vd l Vb / A1 B1 / 1 ~1 \
= . . I +
a b ~ O Bo / ~oO ~0 The first bracketed term goes to zero because the surface is the same and the surface functions are identical. Thè extracted color labeling data remains:
Va' Vb 1 ~1 Va Vb ~0 ~0 This technique is more ~eneral than that given previously, calculatin~
the ratio of the difference of the sensor signais to the sum of these signals, because the latter assumes equal wavelength band powers.
The above is ~ood for equal or unequal wavelength band powers.
v ~
Referring to FIG. 14a, ~n~s~b~e and IR band detector video signals Va and Yb, in digital form, are fed to a detector array noise removal circuit 48 and hence to a memory bus 49. Both detector arrays 41 and 43 are scanned electrically at a high rate and the serialized data on both lines is continuously fed to color pattern decoupling circuitry 45. The data for five points or pixels is read out in parallel to four operators 50-53. All four operator circuits are identical and are implemented with a large scale integrated multiplier-accumulator chip (such as TRW Type TDC1009J) and a memory to store a set of coefficients for each of the five points.
Any given section of the cosine wave is approximated by a second order polynomial. The operators calculate the polynomial which is least squares fitted to a number of points about a given pixel.
Points 1-S in FI6. 15 are typical of the data that is sensed.
The problem to determine the cosine wave 54 that is the best least squares fit, i.e., the sum of the squares of the differences between the curve and each data point is a mininlum. In each 4~34 ~n operator 50-53, the stored coefficients are sequentially multiplied by the sensed pixel data and the sum for all five points is calculated With reference to the normalization equation (Va'/Va ^ Vb'/Vb), operators IX and rX calculate the derivatives of the polynomial at any given point, and operators I8 and IB calculate the values of the polyncmial at any given point. Circuits 55 and 56 perform the indicated divisions and subtractor circuit 57 generates the normalized signal ~n. These operators are applied at each pixel position sucessively in a serial raster scan fashion. The resulting values of TB ~ IB ~ Ix , and Ix are the desired function (value and derivative) of the least squares polynomial fit (of preselected order) about the pixel of evaluation.
Assuming that the surface being measured is flat, the expected normalized signal shown in dashed lines in FIG. 16 is a periodic cosine wave and the periods are equal to one another. A change in the profile, ~z, leads to a phase shift of the detected cosine wave which is shown in full lines. The phase shift at the peak of the cosine wave, and also at the zero, is proportional to the depth or profile change, ~z. The symbols P+ and P , standing for positive and negative peaks of the wave, and Z and Z , standing for positive slope and negative slope zeros of the cosine wave, are indicated in FIG. 16. The normalized signal, it is re-emphasized, is independent of surface reflectivity and roughness variations.
The mathematical basis for peak detection will be explained with reference to FIG. 17. Two oppositely sloped curve sides S1 and S2 meet at peak point xO. The following three functions tend to go to zero at the peak and their sum is a minimum at the peak, Fpeak - > 0.
F1: aa (P(xO`)) ~
At the peak the top is flat and the partial derivative of the polymonial P approximately at xO goes to zero.
11~;4~9~
,~D-;,3 F2: P(S1(xO)) - P(S2(xO)) -- ~ 0 The polynomial approximation of both sides S1 and S2 predicts the same point.
F3: Slope (P(S1)) + Slope (P(S2)) - ~
S The slope of both sides are approximately equal and opposite. The peak is located when the sum of the absolute value of these three functions, taken at many points xO, is minimum.
Turning to FIG. 14b, the normalized signal, although smoothed by the pattern decoupling process, still has localized fluctations and irregularities and may even in some cases contain false zeros and peaks. The normalized signal is fed to a memory buss 60, and sets of nine data points are read out in sequence and presented to a peak and zero detector 61. Linear operators 62-64 are identical to circuits 50-54; they respectively calculate the functions F1, F2, and F3, and their absolute sum is calculated in circuit 65 and passed to the minimum detector 66 by selector 67 which determines whether a peak or zero is being detected. This determination is made by comparing the output of the peak detector components 62-65 and zero detector components 69 and 70 and selecting the signal ~lith the smaller amplitude. This signal is then tracked until it achieves a minimum value at which point the peak or ZerQ is said to exist. When a positive or negative peak in the normalized signal is recognized, the index number of the photodioae array element (1-512) is read out of memory 68.
A set of five d3ta points is also presented to operator 69, which is identical to operators 50 and 52. This operator, I8, determines the cosine wave that is the best least squdres fit.
Absolute value circuit 70 is essentially a rectifier and sends an -lfi-Dl~
output to minilllulll(lctector 66 when a z~ro in th~ nor~ndlized signdl is identified. The index numbers corresponding to the signal zeros ire read out of memory 68. Interpolator 71 is in parallel with peak and zero detector 61, which determines the gross location of the peaks and zeros. This interpolator subdivides the phase shift pixel interval into 32 parts and gets the peaks or zeros between pixels.
The added increment ~Zero is calculated by dividing the normalized - signal, ~n' by the first derivative, Ynl; the added increment ~Peak is calculated by dividing Vn by Ynll, the second derivative. The set of nine data points is fed in parallel to circùits 72-74, which are identical to operator 69. Operator IB yields the best least squares fit cosine wave at a given point, operator IX yields the first derivative of the polynomial at the given point, and operator IXx calculates the second derivative at that point. The indicated divisions are performed by blocks 75 and 76, and the QZero or ~Peak increment is gated by a selector 77.
A classifier 78 identifies the four cosine wave classes or ~eatures P , P , Z , and Z (see FIG. 16) and the one that is selected depends on the sign of IB and Ix. The class information is fed to selector 77 and either the interpolated zero or interpolated peak data is passed to a profile mapping subsystem 79. Profile mapping circuitry 79 has a set of memories such that, given the class, a particular mcmory is accessed and the depth or profile data, z, corresponding to the phase shift can be read out. The need for four memories, one for each of the classes, Z , P , Z , and P , can be seen in FIG. 18 where it is shown that the four classes overlap. In every one of the four hprizontal sets of ramps, each of which begins where the preceding one stops, the number of ramps corresponds to the number o~ periods in the conlplelllentary color pattern. Only two are shown in order to conserve space. There is a known relationship o~
. .
~D-12319 betwe~l) phdse shift an-~ nl~asure~ ~epth, and yiven the phase shift for a particuldr cldss and period, the depth or profile data is stored in this memory. The index number (see FIG. 14b) is also transmitted to a look-up table 80, and corresponding x position data is read out~
This position is simply the location of the given label relative to a fixed frame of reference such as the label transmitter. A complete identification of the profile or depth data requires values of both x and z.
A more general method of reconstructing the surface can be implemented by defining a more refined label structure based on the shape of the original light signal. In specific, if the shape of the normalized sisnal is cosinusoidal, then the phase shift may be reconstructed at each pixel in the signal by simply determining the shift of the corresponding point in a reference signal. This process is limited by the noise in the signal normalization process which is introduced by three sources: the implementation hardware, the signal model, and the optics. The normalization scheme shown above can provide an adequate signal for reconstruction on the pixel level. Th;e peak and zero detector scheme is replaced by a suitable scaling and - 20 correlation circuit to match the normalized signal with a reference signal acquired from a flat surface (z=0). This correlation requires the matching of amplitudes as a function of index and, thus, is somewhat limited at the peak of the signals since the slope is low.
A combination of the Feak detector scheme with the correlation - 25 scheme provides enhanced response in these regions at the expense of some smoothing.
In conclusion, an improved concept for the non-contact measurement of surface profile has been disc!osed. A novel aspcct is the specific approach which allows the determination of profile at all points in the field of YieW. In dddition, these medsurelllents ~i64094 ``;
r~
are relatively imnlune to surface reflectivity and roughness. The color illwllindtion pattern may be composed of ultraviolet, visible, or infrared light.
While the invention has been particularly shown and described with reference to preferred embodiments thereof, it will be understood by those skilled in the art that the foregoing and other changes in form and details may be made therein without departing from the spirit and scope of the invention.
1 9- '
Claims (29)
1. The method of measuring surface profile comprising the steps of:
scanning an object with an optical transmitter that projects onto the surface of said object a color illumination pattern comprised of at least two different wavelengths of light;
sensing reflected light at a known parallax angle and separately detecting the different wavelengths which shift position on a detector array depending upon the depth from the object surface to a reference, and generating separate electrical sensor signals corresponding to the different wavelength detected light intensities; and processing said sensor signals to extract depth data from which the surface profile can be determined.
scanning an object with an optical transmitter that projects onto the surface of said object a color illumination pattern comprised of at least two different wavelengths of light;
sensing reflected light at a known parallax angle and separately detecting the different wavelengths which shift position on a detector array depending upon the depth from the object surface to a reference, and generating separate electrical sensor signals corresponding to the different wavelength detected light intensities; and processing said sensor signals to extract depth data from which the surface profile can be determined.
2. The method of claim 1 wherein said color pattern is projected such that every wavelength power incident on said surface continuously varies.
3. The method of claim 1 wherein said color pattern is projected such that the relative amounts of every wavelength power incident on said surface continuously varies whereby the position of each point in the projected color pattern can be identified.
4. The method of claim 3 wherein the total wavelength power is approximately constant at every point in the projected pattern.
5. The method of claim 3 wherein each wavelength power varies periodically.
6. The method of claim 1 wherein said color pattern is a complementary pattern composed of only two wavelengths or wavelength bands of light.
7. The method of claim 1 wherein the processing of said sensor signals includes deriving a normalized signal which is substantially independent of surface reflectivity and roughness variations.
8. The method of claim 7 wherein the signal processing further includes determining the phase shift between said normalized signal and a reference and obtaining said depth data from the phase shift.
9. The method of claim 8 wherein the phase shift is determined by locating the peaks and zeros of said normalized signal.
10. The method of measuring surface profile comprising the steps of:
projecting onto the surface of an object a color illumination pattern comprised of at least two different wavelengths bands of tight that each have a continuous variation of power;
sensing reflected light at a given parallax angle and extracting and separately detecting said wavelength bands which shift position depending upon the depth from the object surface to a reference, and generating separate electrical sensor signals corresponding to the detected light intensity of each wavelength band;
scanning the surface of said object with the projected color pattern while generating said sensor signals; and processing said sensor signals to extract a normalized signal that is substantially independent of surface reflectivity and roughness variations, and processing said normalized signal to yield depth data from which the surface profile can be determined.
projecting onto the surface of an object a color illumination pattern comprised of at least two different wavelengths bands of tight that each have a continuous variation of power;
sensing reflected light at a given parallax angle and extracting and separately detecting said wavelength bands which shift position depending upon the depth from the object surface to a reference, and generating separate electrical sensor signals corresponding to the detected light intensity of each wavelength band;
scanning the surface of said object with the projected color pattern while generating said sensor signals; and processing said sensor signals to extract a normalized signal that is substantially independent of surface reflectivity and roughness variations, and processing said normalized signal to yield depth data from which the surface profile can be determined.
11. The method of claim 10 wherein the projected color pattern is a complementary color pattern composed of two wavelength bands of light.
12. The method of claim 11 wherein said color pattern is projected at a parallax angle to maximize the sensitivity to profile changes while at the same time optimizing variations in reflected light as a function of surface roughness.
13. The method of claim 11 wherein said color pattern is produced by extracting said wavelength bands by optical filtering of the emitted wavelengths of a multispectral light source, and recombining said wavelength bands by the use of a spatial pattern.
14. The method of measuring surface profile comprising the steps of:
scanning an object with an optical transmitter that projects onto the surface of said object a complementary color pattern composed of two different wavelength bands of light that combine periodically to have a continuous variation of power ratios;
sensing reflected light with an optical receiver at a given parallax angle and extracting and separately detecting said wavelength bands which shift position on a pair of linear detector arrays, one for each wavelength band, depending upon the depth from the object surface to a reference, and generating first and second electrical sensor signals each corresponding to the respective wavelength band detected light intensity; and processing said first and second sensor signals to derive a normalized signal which is substantially independent of surface reflectivity and roughness variations, and processing said normalized signal to yield depth data from which the surface profile can be determined.
scanning an object with an optical transmitter that projects onto the surface of said object a complementary color pattern composed of two different wavelength bands of light that combine periodically to have a continuous variation of power ratios;
sensing reflected light with an optical receiver at a given parallax angle and extracting and separately detecting said wavelength bands which shift position on a pair of linear detector arrays, one for each wavelength band, depending upon the depth from the object surface to a reference, and generating first and second electrical sensor signals each corresponding to the respective wavelength band detected light intensity; and processing said first and second sensor signals to derive a normalized signal which is substantially independent of surface reflectivity and roughness variations, and processing said normalized signal to yield depth data from which the surface profile can be determined.
15. The method of claim 14 wherein the projected complementary color pattern is produced by extracting said wavelength bands from a single multispectral light source, said wavelength bands further having approximately equal powers, and recombining said wavelength bands by the use of a spatial pattern.
16. The method of claim 14 wherein said first and second sensor signals are processed by calculating (Va'/Va) - (Vb'/Vb), where Va' and Vb' are the first derivatives of sensor signals Va and Vb, to yield said normalized signal.
17. The method of claim 16 wherein said normalized signal is processed by determining the phase shift between said normalized signal and a reference by locating the peaks and zeros of said normalized signal, and obtaining said depth data from the phase shift.
18. A system for acquiring surface profile measurements comprising:
an optical sensor for scanning the surface of an object;
said sensor comprising an optical transmitter having a first optical system for supplying at least two wavelengths of light which are combined to project a color pattern onto the surface of said object;
said sensor further comprising an optical receiver for sensing reflected light at a predetermined parallax angle and having a second optical system for extracting said wavelengths of light, and that includes a detector array on which said wavelengths shift position depending upon the depth from the surface to a reference, said detector array generating separate electrical sensor signals corresponding to the detected light intensity of each wavelength;
and a signal processor for deriving from said sensor signals depth data from which the surface profile can be mapped.
an optical sensor for scanning the surface of an object;
said sensor comprising an optical transmitter having a first optical system for supplying at least two wavelengths of light which are combined to project a color pattern onto the surface of said object;
said sensor further comprising an optical receiver for sensing reflected light at a predetermined parallax angle and having a second optical system for extracting said wavelengths of light, and that includes a detector array on which said wavelengths shift position depending upon the depth from the surface to a reference, said detector array generating separate electrical sensor signals corresponding to the detected light intensity of each wavelength;
and a signal processor for deriving from said sensor signals depth data from which the surface profile can be mapped.
19. The system of claim 18 wherein said color pattern is projected approximately normal to said surface and said parallax angle is fixed at about 26° to reduce sensitivity to local surface roughness fluctuations.
20. The system of claim 18 wherein said signal processor has color pattern decoupling circuitry for extracting from said sensor signals a normalized signal that is substantially independent of surface reflectivity and roughness variations, and means for processing said normalized signal to yield said depth data.
21. A system for acquiring surface profile measurements comprising:
an optical sensor for scanning the surface of an object;
21. A system for acquiring surface profile measurements comprising:
an optical sensor for scanning the surface of an object;
Claim 21 Cont'd said sensor comprising an optical transmitter having a multispectral source of light, a first subsystem for extracting two approximately equal power wavelength bands from the emitted wavelengths of said light source, and a second subsystem for recombining said wavelength bands and for projecting onto the surface a periodic complementary color pattern;
said sensor further comprising an optical receiver for sensing reflected light at a predetermined parallax angle and having a third subsystem for separating said wavelength bands and a pair of detector arrays on which said wavelength bands shift position depending upon the depth from the object surface to a reference, said detector arrays generating a pair of electrical sensor signals corresponding to the detected light intensities of the two wavelength bands; and a signal processor comprising color pattern decoupling circuitry for processing said sensor signals and extracting a normalized signal that is substantially independent of surface reflectivity and roughness variations, and means for processing said normalized signal to yield the phase shift between said normalized signal and a reference and hence depth data from which the surface profile can be mapped.
22. The system of claim 21 wherein said first sub-system includes a set of dichroic mirrors and said second subsystem includes a condensing lens for each wavelength band and a patterned mirror which transmits
said sensor further comprising an optical receiver for sensing reflected light at a predetermined parallax angle and having a third subsystem for separating said wavelength bands and a pair of detector arrays on which said wavelength bands shift position depending upon the depth from the object surface to a reference, said detector arrays generating a pair of electrical sensor signals corresponding to the detected light intensities of the two wavelength bands; and a signal processor comprising color pattern decoupling circuitry for processing said sensor signals and extracting a normalized signal that is substantially independent of surface reflectivity and roughness variations, and means for processing said normalized signal to yield the phase shift between said normalized signal and a reference and hence depth data from which the surface profile can be mapped.
22. The system of claim 21 wherein said first sub-system includes a set of dichroic mirrors and said second subsystem includes a condensing lens for each wavelength band and a patterned mirror which transmits
Claim 22 Cont'd one wavelength band and reflects the other.
23. The system of claim 22 wherein said patterned mirror has a bar pattern of reflective material.
24. The system of claim 23 wherein said light source is an incandescent tungsten lamp and said wavelength bands are in the visible and infrared regions.
25. The system of claim 21 wherein said third subsystem includes a dichroic mirror and pass filters and said detector arrays, one for each wavelength band, are liner photodiode arrays.
26. The system of claim 21 wherein said pattern decoupling circuitry calculates (Va'/Va-Vb'/Vb), where Va' and Vb' are the first derivatives of sensor signals Va and Vb, to yield said normalized signal.
27. The system of claim 26 wherein said means for processing said normalized signal comprises a peak and zero detector circuit to realize the phase in each period between said normalized signal and the reference and with which is associated an interpolator.
28. The system of claim 27 wherein said peak detector calculates three functions for a curve with two sides Sl and S2 that meet at xO that tend to be zero when there is a peak, sums the absolute values of the three functions, and produces an output when a minimum sum is detected.
29. The system of claim 28 wherein said functions are that the derivative of a polynomial approximation at xO
goes to zero, that the polynominal approximation of both sides predicts the same point, and that the slopes are approximately equal in magnitude and opposite in sign.
goes to zero, that the polynominal approximation of both sides predicts the same point, and that the slopes are approximately equal in magnitude and opposite in sign.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| US06/158,372 US4349277A (en) | 1980-06-11 | 1980-06-11 | Non-contact measurement of surface profile |
| US158,372 | 1980-06-11 |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CA1164094A true CA1164094A (en) | 1984-03-20 |
Family
ID=22567820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CA000379180A Expired CA1164094A (en) | 1980-06-11 | 1981-06-05 | Non-contact measurement of surface profile |
Country Status (9)
| Country | Link |
|---|---|
| US (1) | US4349277A (en) |
| JP (1) | JPS5724810A (en) |
| CA (1) | CA1164094A (en) |
| DE (1) | DE3122712A1 (en) |
| FR (1) | FR2484633B1 (en) |
| GB (1) | GB2078944B (en) |
| IL (1) | IL62959A (en) |
| IT (1) | IT1139362B (en) |
| NL (1) | NL8102813A (en) |
Families Citing this family (114)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE3170315D1 (en) * | 1981-10-09 | 1985-06-05 | Ibm Deutschland | Interpolating light section process |
| DE3233013A1 (en) * | 1982-09-06 | 1984-03-08 | Siemens AG, 1000 Berlin und 8000 München | OPTICAL ARRANGEMENT FOR DETECTING AND EVALUATING THE LOCATION OF AN OBJECT |
| US4520388A (en) * | 1982-11-01 | 1985-05-28 | General Electric Company | Optical signal projector |
| US4491719A (en) * | 1982-12-20 | 1985-01-01 | General Electric Company | Light pattern projector especially for welding |
| US4687324A (en) * | 1983-05-18 | 1987-08-18 | Gerd Selbach | Laser-based measuring system |
| NL8302228A (en) * | 1983-06-22 | 1985-01-16 | Optische Ind De Oude Delft Nv | MEASURING SYSTEM FOR USING A TRIANGULAR PRINCIPLE, CONTACT-FREE MEASURING A DISTANCE GIVEN BY A SURFACE CONTOUR TO AN OBJECTIVE LEVEL. |
| US4692690A (en) * | 1983-12-26 | 1987-09-08 | Hitachi, Ltd. | Pattern detecting apparatus |
| DE3413605A1 (en) * | 1984-04-11 | 1985-10-17 | Fraunhofer-Gesellschaft zur Förderung der angewandten Forschung e.V., 8000 München | Optical method for measuring the profile of surfaces with a locally strongly fluctuating reflection factor |
| US4650333A (en) * | 1984-04-12 | 1987-03-17 | International Business Machines Corporation | System for measuring and detecting printed circuit wiring defects |
| JPS60220808A (en) * | 1984-04-17 | 1985-11-05 | Inax Corp | Surface inspecting method |
| US4664470A (en) * | 1984-06-21 | 1987-05-12 | General Electric Company | Method and system for structured radiation production including a composite filter and method of making |
| DE3428718A1 (en) * | 1984-08-03 | 1986-02-06 | Erwin Sick Gmbh Optik-Elektronik, 7808 Waldkirch | Optical instrument for determining the waviness of material surfaces |
| US4641972A (en) * | 1984-09-14 | 1987-02-10 | New York Institute Of Technology | Method and apparatus for surface profilometry |
| US4657394A (en) * | 1984-09-14 | 1987-04-14 | New York Institute Of Technology | Apparatus and method for obtaining three dimensional surface contours |
| US4664514A (en) * | 1984-11-06 | 1987-05-12 | General Electric Company | Method of enhancing surface features and detecting same |
| US4634879A (en) * | 1985-03-21 | 1987-01-06 | General Electric Company | Method and system for determining surface profile information |
| US4687325A (en) * | 1985-03-28 | 1987-08-18 | General Electric Company | Three-dimensional range camera |
| US4658368A (en) * | 1985-04-30 | 1987-04-14 | Canadian Patents And Development Limited-Societe Canadienne Des Brevets Et D'exploitation Limitee | Peak position detector |
| US4645917A (en) * | 1985-05-31 | 1987-02-24 | General Electric Company | Swept aperture flying spot profiler |
| FR2589242B1 (en) * | 1985-10-25 | 1988-11-25 | Oreal | PROCESS FOR EXAMINING THE SURFACE OF A SAMPLE AND APPARATUS FOR IMPLEMENTING SAME |
| US4687326A (en) * | 1985-11-12 | 1987-08-18 | General Electric Company | Integrated range and luminance camera |
| GB8615197D0 (en) * | 1986-06-21 | 1986-07-23 | Renishaw Plc | Opto-electronic scale reading apparatus |
| JPH0615968B2 (en) * | 1986-08-11 | 1994-03-02 | 伍良 松本 | Three-dimensional shape measuring device |
| US4868404A (en) * | 1987-04-23 | 1989-09-19 | Hajime Industries, Ltd. | Surface inspection apparatus using a mask system to monitor uneven surfaces |
| US4850712A (en) * | 1988-02-01 | 1989-07-25 | Caterpillar Inc. | Method and system for determining surface profile information |
| US4946275A (en) * | 1988-03-31 | 1990-08-07 | General Dynamics Corporation Convair Division | Fiber optic position transducer |
| US5011960A (en) * | 1988-05-20 | 1991-04-30 | Fujitsu Limited | Wiring pattern detection method and apparatus |
| DE3907430C1 (en) * | 1988-12-23 | 1991-03-21 | Klaus 8206 Bruckmuehl De Pfister | |
| US4981360A (en) * | 1989-05-10 | 1991-01-01 | Grumman Aerospace Corporation | Apparatus and method for projection moire mapping |
| US5025285A (en) * | 1989-05-10 | 1991-06-18 | Grumman Aerospace Corporation | Apparatus and method for shadow moire mapping |
| US5169194A (en) * | 1989-07-12 | 1992-12-08 | Kyokuto Kaihatsu Kogyo Co., Ltd. | Container carrier |
| DE3938714A1 (en) * | 1989-11-23 | 1991-05-29 | Bernd Dr Breuckmann | Optical determination of object shapes, shape variations - using structured coloured light projected onto objects for high resolution, dynamic measurement |
| US5224049A (en) * | 1990-04-10 | 1993-06-29 | Mushabac David R | Method, system and mold assembly for use in preparing a dental prosthesis |
| US5343391A (en) * | 1990-04-10 | 1994-08-30 | Mushabac David R | Device for obtaining three dimensional contour data and for operating on a patient and related method |
| US5257184A (en) * | 1990-04-10 | 1993-10-26 | Mushabac David R | Method and apparatus with multiple data input stylii for collecting curvilinear contour data |
| US5347454A (en) * | 1990-04-10 | 1994-09-13 | Mushabac David R | Method, system and mold assembly for use in preparing a dental restoration |
| US5569578A (en) * | 1990-04-10 | 1996-10-29 | Mushabac; David R. | Method and apparatus for effecting change in shape of pre-existing object |
| US5545039A (en) * | 1990-04-10 | 1996-08-13 | Mushabac; David R. | Method and apparatus for preparing tooth or modifying dental restoration |
| US5562448A (en) * | 1990-04-10 | 1996-10-08 | Mushabac; David R. | Method for facilitating dental diagnosis and treatment |
| CA2017518A1 (en) * | 1990-05-24 | 1991-11-24 | Her Majesty The Queen, In Right Of Canada, As Represented By The Ministe R Of The National Research Council Of Canada | Colour-range imaging |
| US5134303A (en) * | 1990-08-14 | 1992-07-28 | Flexus, Inc. | Laser apparatus and method for measuring stress in a thin film using multiple wavelengths |
| US5248889A (en) * | 1990-08-14 | 1993-09-28 | Tencor Instruments, Inc. | Laser apparatus and method for measuring stress in a thin film using multiple wavelengths |
| DE69130147T2 (en) * | 1990-10-03 | 1999-04-01 | Aisin Seiki | Automatic control system for lateral guidance |
| US5390118A (en) * | 1990-10-03 | 1995-02-14 | Aisin Seiki Kabushiki Kaisha | Automatic lateral guidance control system |
| US5202742A (en) * | 1990-10-03 | 1993-04-13 | Aisin Seiki Kabushiki Kaisha | Laser radar for a vehicle lateral guidance system |
| DE4134546A1 (en) * | 1991-09-26 | 1993-04-08 | Steinbichler Hans | METHOD AND DEVICE FOR DETERMINING THE ABSOLUTE COORDINATES OF AN OBJECT |
| US5636025A (en) * | 1992-04-23 | 1997-06-03 | Medar, Inc. | System for optically measuring the surface contour of a part using more fringe techniques |
| US5461226A (en) * | 1993-10-29 | 1995-10-24 | Loral Infrared & Imaging Systems, Inc. | Photon counting ultraviolet spatial image sensor with microchannel photomultiplying plates |
| US7218448B1 (en) * | 1997-03-17 | 2007-05-15 | The Regents Of The University Of Colorado | Extended depth of field optical systems |
| US20020118457A1 (en) * | 2000-12-22 | 2002-08-29 | Dowski Edward Raymond | Wavefront coded imaging systems |
| US6911638B2 (en) | 1995-02-03 | 2005-06-28 | The Regents Of The University Of Colorado, A Body Corporate | Wavefront coding zoom lens imaging systems |
| US20020195548A1 (en) * | 2001-06-06 | 2002-12-26 | Dowski Edward Raymond | Wavefront coding interference contrast imaging systems |
| JPH0961132A (en) * | 1995-08-28 | 1997-03-07 | Olympus Optical Co Ltd | Three-dimensional-shape measuring apparatus |
| DE19536294C2 (en) * | 1995-09-29 | 2003-12-18 | Daimler Chrysler Ag | Method for geometric navigation of 3D optical sensors for the three-dimensional measurement of objects |
| US6229619B1 (en) | 1996-02-12 | 2001-05-08 | Massachusetts Institute Of Technology | Compensation for measurement uncertainty due to atmospheric effects |
| US6690474B1 (en) | 1996-02-12 | 2004-02-10 | Massachusetts Institute Of Technology | Apparatus and methods for surface contour measurement |
| US6031612A (en) * | 1996-02-12 | 2000-02-29 | Massachusetts Institute Of Technology | Apparatus and methods for contour measurement using movable sources |
| US5708498A (en) * | 1996-03-04 | 1998-01-13 | National Research Council Of Canada | Three dimensional color imaging |
| FR2748322A1 (en) * | 1996-05-02 | 1997-11-07 | Cohen Sabban Joseph | Opto-electronic instrument for digitising surface of object |
| JP3417222B2 (en) * | 1996-08-07 | 2003-06-16 | 松下電器産業株式会社 | Real-time range finder |
| FR2758076A1 (en) * | 1997-01-09 | 1998-07-10 | Sabban Joseph Cohen | Opto-electronic system for three=dimension digitising of teeth dental data without contact and in real=time |
| US6252623B1 (en) | 1998-05-15 | 2001-06-26 | 3Dmetrics, Incorporated | Three dimensional imaging system |
| AU3994799A (en) * | 1999-05-14 | 2000-12-05 | 3Dmetrics, Incorporated | Color structured light 3d-imaging system |
| JP2001174409A (en) * | 1999-12-15 | 2001-06-29 | Internatl Business Mach Corp <Ibm> | Two-wavelength tube, lighting apparatus for inspection, inspection apparatus and inspection method |
| US6888640B2 (en) | 2000-02-04 | 2005-05-03 | Mario J. Spina | Body spatial dimension mapper |
| US6639685B1 (en) | 2000-02-25 | 2003-10-28 | General Motors Corporation | Image processing method using phase-shifted fringe patterns and curve fitting |
| EP1306646A4 (en) * | 2000-06-07 | 2006-08-16 | Citizen Watch Co Ltd | Lattice pattern projector using liquid crystal lattice |
| US6417950B1 (en) * | 2000-08-28 | 2002-07-09 | University Technology Corporation | Three-color imaging on each pixel for increased resolution |
| US6536898B1 (en) * | 2000-09-15 | 2003-03-25 | The Regents Of The University Of Colorado | Extended depth of field optics for human vision |
| US6369879B1 (en) | 2000-10-24 | 2002-04-09 | The Regents Of The University Of California | Method and apparatus for determining the coordinates of an object |
| US6504605B1 (en) | 2000-10-24 | 2003-01-07 | The Regents Of The University Of California | Method and apparatus for determining the coordinates of an object |
| US6873733B2 (en) | 2001-01-19 | 2005-03-29 | The Regents Of The University Of Colorado | Combined wavefront coding and amplitude contrast imaging systems |
| US6842297B2 (en) | 2001-08-31 | 2005-01-11 | Cdm Optics, Inc. | Wavefront coding optics |
| US20030067537A1 (en) * | 2001-10-04 | 2003-04-10 | Myers Kenneth J. | System and method for three-dimensional data acquisition |
| US20050110868A1 (en) * | 2001-10-04 | 2005-05-26 | Myers Kenneth J. | System and method for inputting contours of a three-dimensional subject to a computer |
| DE10205132A1 (en) * | 2002-02-07 | 2003-08-28 | Bfi Vdeh Inst Angewandte Forschung Gmbh | Method and device for the optical measurement of the surface shape and for the optical surface inspection of moving strips in rolling and further processing plants |
| DE10212364A1 (en) * | 2002-03-20 | 2003-10-16 | Steinbichler Optotechnik Gmbh | Method and device for determining the absolute coordinates of an object |
| US6700668B2 (en) * | 2002-06-25 | 2004-03-02 | General Electric Company | Method of measuring a part with a wide range of surface reflectivities |
| US7103212B2 (en) * | 2002-11-22 | 2006-09-05 | Strider Labs, Inc. | Acquisition of three-dimensional images by an active stereo technique using locally unique patterns |
| US6969821B2 (en) * | 2003-06-30 | 2005-11-29 | General Electric Company | Airfoil qualification system and method |
| DE602005009432D1 (en) | 2004-06-17 | 2008-10-16 | Cadent Ltd | Method and apparatus for color forming a three-dimensional structure |
| US7433058B2 (en) * | 2004-07-12 | 2008-10-07 | Solvision Inc. | System and method for simultaneous 3D height measurements on multiple sides of an object |
| US7646896B2 (en) * | 2005-08-02 | 2010-01-12 | A4Vision | Apparatus and method for performing enrollment of user biometric information |
| WO2006031143A1 (en) * | 2004-08-12 | 2006-03-23 | A4 Vision S.A. | Device for contactlessly controlling the surface profile of objects |
| AU2005285558C1 (en) * | 2004-08-12 | 2012-05-24 | A4 Vision S.A | Device for biometrically controlling a face surface |
| US7436524B2 (en) * | 2004-11-26 | 2008-10-14 | Olympus Corporation | Apparatus and method for three-dimensional measurement and program for allowing computer to execute method for three-dimensional measurement |
| DE102005014525B4 (en) * | 2005-03-30 | 2009-04-16 | Siemens Ag | Device for determining spatial coordinates of object surfaces |
| JP2007114071A (en) | 2005-10-20 | 2007-05-10 | Omron Corp | Three-dimensional shape measuring apparatus, program, computer-readable storage medium, and three-dimensional shape measuring method |
| US7576347B2 (en) * | 2005-10-24 | 2009-08-18 | General Electric Company | Method and apparatus for optically inspecting an object using a light source |
| US20070090310A1 (en) * | 2005-10-24 | 2007-04-26 | General Electric Company | Methods and apparatus for inspecting an object |
| CA2528791A1 (en) * | 2005-12-01 | 2007-06-01 | Peirong Jia | Full-field three-dimensional measurement method |
| JP4801457B2 (en) * | 2006-02-02 | 2011-10-26 | 株式会社リコー | Surface defect inspection apparatus, surface defect inspection method, and surface defect inspection program |
| WO2008044096A1 (en) * | 2006-10-13 | 2008-04-17 | Yeditepe Üniversitesi | Method for three-dimensionally structured light scanning of shiny or specular objects |
| CA2597891A1 (en) * | 2007-08-20 | 2009-02-20 | Marc Miousset | Multi-beam optical probe and system for dimensional measurement |
| FI20075975L (en) * | 2007-12-31 | 2009-07-01 | Metso Automation Oy | Measuring a trajectory |
| US8094321B2 (en) * | 2008-02-26 | 2012-01-10 | Caterpillar Inc. | Photogrammetric target and related method |
| JP2009300263A (en) * | 2008-06-13 | 2009-12-24 | Mitsutoyo Corp | Two-wavelength laser interferometer and method of adjusting optical axis in the same |
| EP2159538A1 (en) | 2008-08-28 | 2010-03-03 | Omnitron AG | Method and device for calculating the location and/or shape of an object |
| US8723118B2 (en) * | 2009-10-01 | 2014-05-13 | Microsoft Corporation | Imager for constructing color and depth images |
| WO2011054083A1 (en) | 2009-11-04 | 2011-05-12 | Technologies Numetrix Inc. | Device and method for obtaining three-dimensional object surface data |
| KR101207198B1 (en) | 2010-01-18 | 2012-12-03 | 주식회사 고영테크놀러지 | Board inspection apparatus |
| DE102010030435A1 (en) * | 2010-06-23 | 2011-12-29 | Carl Zeiss Smt Gmbh | metrology system |
| US8760499B2 (en) * | 2011-04-29 | 2014-06-24 | Austin Russell | Three-dimensional imager and projection device |
| CN102679908A (en) * | 2012-05-10 | 2012-09-19 | 天津大学 | Dynamic measurement method of three-dimensional shape projected by dual-wavelength fiber interference fringe |
| US9175957B2 (en) * | 2012-09-24 | 2015-11-03 | Alces Technology, Inc. | Grayscale patterns from binary spatial light modulators |
| DE102012021185A1 (en) | 2012-10-30 | 2014-04-30 | Smart Optics Sensortechnik Gmbh | Method for 3D optical measurement of teeth with reduced point-spread function |
| US10412280B2 (en) | 2016-02-10 | 2019-09-10 | Microsoft Technology Licensing, Llc | Camera with light valve over sensor array |
| DE102018115673A1 (en) * | 2018-06-28 | 2020-02-13 | Carl Zeiss Ag | Methods and devices for pattern projection |
| DE102018005506B4 (en) * | 2018-07-12 | 2021-03-18 | Wenzel Group GmbH & Co. KG | Optical sensor system for a coordinate measuring machine, method for detecting a measuring point on a surface of a measuring object and coordinate measuring machine |
| CN112740111A (en) * | 2018-09-21 | 2021-04-30 | Asml荷兰有限公司 | Radiation system |
| WO2020093321A1 (en) * | 2018-11-08 | 2020-05-14 | 成都频泰鼎丰企业管理中心（有限合伙） | Three-dimensional measurement device |
| DE102020008179B4 (en) | 2020-10-22 | 2023-10-26 | Smart Optics Sensortechnik Gmbh | Method and device for optical three-dimensional measurement of objects |
| DE102020127894B4 (en) | 2020-10-22 | 2022-09-22 | Smart Optics Sensortechnik Gmbh | Method and device for the optical three-dimensional measurement of objects |
| JP7370355B2 (en) * | 2021-03-25 | 2023-10-27 | ジャパンマリンユナイテッド株式会社 | Evaluation method and device for roughening metal surfaces |
Family Cites Families (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| DE911662C (en) * | 1941-11-06 | 1954-05-17 | Emil Busch A G | Microscope for visualizing surface and layer structures |
| FR1312164A (en) * | 1960-12-22 | 1962-12-14 | United States Steel Corp | Method and apparatus for determining the profile of a surface |
| DE1473780A1 (en) * | 1965-12-30 | 1969-03-13 | Bbc Brown Boveri & Cie | Device for non-contact measurement of contours |
| US3619065A (en) * | 1967-07-29 | 1971-11-09 | Kenneth Leslie Agnew | Verification of three-dimensional shapes by means of contour lines |
| JPS5418145B1 (en) * | 1971-06-08 | 1979-07-05 | ||
| JPS5833386B2 (en) * | 1975-01-08 | 1983-07-19 | 株式会社デンソー | Air fuel ratio control device |
| US4175862A (en) * | 1975-08-27 | 1979-11-27 | Solid Photography Inc. | Arrangement for sensing the geometric characteristics of an object |
| US4227813A (en) * | 1977-03-10 | 1980-10-14 | Centre De Recherches Metallurgiques Centrum Voor Research In De Metallurgie | Process for determining a dimension of an object |
| US4158507A (en) * | 1977-07-27 | 1979-06-19 | Recognition Equipment Incorporated | Laser measuring system for inspection |
| US4180329A (en) * | 1978-03-23 | 1979-12-25 | The United States Of America As Represented By The Secretary Of The Air Force | Single blade proximity probe |
-
1980
- 1980-06-11 US US06/158,372 patent/US4349277A/en not_active Expired - Lifetime
-
1981
- 1981-05-26 IL IL62959A patent/IL62959A/en unknown
- 1981-06-01 GB GB8116660A patent/GB2078944B/en not_active Expired
- 1981-06-05 CA CA000379180A patent/CA1164094A/en not_active Expired
- 1981-06-06 DE DE19813122712 patent/DE3122712A1/en not_active Ceased
- 1981-06-08 IT IT22186/81A patent/IT1139362B/en active
- 1981-06-10 JP JP8831181A patent/JPS5724810A/en active Granted
- 1981-06-11 FR FR8111491A patent/FR2484633B1/en not_active Expired
- 1981-06-11 NL NL8102813A patent/NL8102813A/en not_active Application Discontinuation
Also Published As
| Publication number | Publication date |
|---|---|
| FR2484633B1 (en) | 1987-01-09 |
| GB2078944A (en) | 1982-01-13 |
| DE3122712A1 (en) | 1982-03-18 |
| IT8122186A0 (en) | 1981-06-08 |
| GB2078944B (en) | 1984-03-28 |
| IL62959A0 (en) | 1981-07-31 |
| JPS5724810A (en) | 1982-02-09 |
| IL62959A (en) | 1984-07-31 |
| US4349277A (en) | 1982-09-14 |
| JPH0330802B2 (en) | 1991-05-01 |
| FR2484633A1 (en) | 1981-12-18 |
| NL8102813A (en) | 1982-01-04 |
| IT1139362B (en) | 1986-09-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CA1164094A (en) | Non-contact measurement of surface profile | |
| US5365340A (en) | Apparatus and method for measuring the thickness of thin films | |
| EP0545738B1 (en) | Apparatus for measuring the thickness of thin films | |
| US5675407A (en) | Color ranging method for high speed low-cost three dimensional surface profile measurement | |
| US5293214A (en) | Apparatus and method for performing thin film layer thickness metrology by deforming a thin film layer into a reflective condenser | |
| EP0577399B1 (en) | Apparatus and method for performing thin film layer thickness metrology on a thin film layer having shape deformations and local slope variations | |
| JP6161714B2 (en) | Method for controlling the linear dimension of a three-dimensional object | |
| CA2310432C (en) | Three dimensional optical scanning | |
| JPH0153721B2 (en) | ||
| JP2004537732A (en) | Three-dimensional imaging by projecting interference fringes and evaluating absolute phase mapping | |
| US4725146A (en) | Method and apparatus for sensing position | |
| RU125335U1 (en) | DEVICE FOR MONITORING LINEAR SIZES OF THREE-DIMENSIONAL OBJECTS | |
| US5408323A (en) | Position fixing apparatus with radiation sensor | |
| EP0458283A1 (en) | Distance information obtaining device | |
| JPH03243804A (en) | Shape measuring method for aspherical surface | |
| Erskine et al. | Imaging white-light VISAR | |
| RU2194256C1 (en) | Autocorrelator of luminous pulses | |
| JPH0758169B2 (en) | Deformation measuring device | |
| JPH0416896Y2 (en) | ||
| Deck et al. | Coherence scanning in a geometrically desensitized interferometer | |
| JPS61130807A (en) | Optical measuring apparatus of microscopical clearance | |
| Clarke et al. | Laser scanning systems for image acquisition | |
| IL119299A (en) | Apparatus and method for measuring the thickness of thin films | |
| Amir et al. | Three-dimensional line-scan intensity ratio sensing |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| MKEX | Expiry |