WO2013020277A1

WO2013020277A1 - Steel ball surface development method and device based on multiple image sensors

Info

Publication number: WO2013020277A1
Application number: PCT/CN2011/078183
Authority: WO
Inventors: 王向军; 栗琳; 王仲; 裴芳莹; 刘新波
Original assignee: 天津大学
Priority date: 2011-08-10
Filing date: 2011-08-10
Publication date: 2013-02-14
Also published as: JP5430801B2; JP2013535017A

Abstract

A steel ball surface unfolding method and device based on multiple image sensors. The method comprises: setting a straight guide rail (5), the straight guide rail (5) having a gradient; disposing two symmetrical image sensors (1) with the same structure at two sides above an axis of the straight guide rail (5); adjusting parameters of the two image sensors (1) to be consistent with each other and adjusting optical parameters of two lenses (2) to be consistent with each other; disposing a steel ball (6) to be measured on the straight guide rail (5) with a groove; the steel ball (6) to be measured performing linear motion on the straight guide rail (5) with the groove by means of gravity; and when the steel ball (6) to be measured passes a position under the image sensors (1), the two image sensors (1) continuously collecting images of the steel ball (6) to be measured. The device comprises straight guide rail (5) having a gradient and formed of a groove at a middle portion of a workbench, and two supports (3) symmetrically disposed at two ends of the workbench and located at two sides of the straight guide rail (5). An image sensor (1) is disposed at the top end of each of the two supports (3), and lenses (2) of the two image sensors both correspond to a central axis at the bottom portion of the straight guide rail (5). The present invention solve the problem that it is difficult to unfold the surface of the steel ball precisely at a high speed.

Description

Method and device for unfolding steel ball surface based on multi-image sensor

The invention relates to a method for unfolding a steel ball surface. In particular, it relates to a method and a device for unfolding a steel ball surface based on a multi-image sensor that can fully cover the surface of the ball by one-dimensional rolling of the steel ball. Background technique

Whether it is a bearing component or an independent rolling element, the steel ball has a wide range of applications. Due to the processing technology and precision, the surface of the steel ball will be worn, cracked, rusted, pitted, etc., which have a significant impact on the speed, noise, vibration, and life of the bearing. Therefore, the detection of surface defects of steel balls is of great significance.

At present, most domestic manufacturers use artificial methods to check the surface quality of steel balls with naked eyes or with low magnification mirrors. However, the artificial observation has strong randomness, especially for the surface defects of the micro steel ball, which are difficult to find, and it is easy to cause missed detection or misdetection. Secondly, the manual pushing can not guarantee the full deployment of the spherical surface, and the missing inspection will occur. The inspection personnel have a long time to observe the smoothness. High steel balls are prone to visual fatigue. Therefore, an automatic detecting device capable of correctly and effectively evaluating the surface quality of a steel ball has become an urgent demand in the bearing industry.

The automatic detection device for the surface quality of the steel ball is developed. The surface expansion device of the steel ball is the core, which is a technical problem that has been difficult to solve at home and abroad. Up to now, there have been many methods for detecting steel ball defects, including eddy current testing technology, ultrasonic flaw detection technology, photoelectric detection technology and visual inspection technology.

In the conventional steel ball surface defect image detecting device, only a single image sensor is usually used. The sensor can only acquire the spherical crown portion of a certain diameter in one image. When the steel ball wraps around the center of the ball and rotates perpendicular to the diameter of the optical axis, even if the sensor is continuously imaged, only an image of an annulus region on the surface of the steel ball can be obtained. In order to obtain a complete image of the surface of the steel ball, the steel ball must have a second dimensional mechanical motion. And this is the result of the randomness and complexity of obtaining a full coverage image of the steel ball surface.

Therefore, the expansion device of the steel ball still has the following problems: 1. Almost all stay in the two-dimensional motion, and the mechanical structure is complicated. For example, the more mature meridional expansion method based on eddy current testing is based on the assumption that the steel ball only performs pure rolling between the unwinding wheel and the driving wheel without causing slip. The unfolding wheel is prone to wear in the application, and the damaged unfolding wheel may cause the steel ball to be unfolded insufficiently, which may cause the defect to be missed, requiring frequent replacement, and high maintenance cost. A similar problem exists in the subsequent development of the warp and weft scanning method. 2, can not achieve full coverage. In the existing method, the full deployment of the steel balls is relatively random. For example, in some vision-based steel ball surface automatic detecting devices, the detecting chamber in the feeding tray carries the steel ball to rotate at a certain speed, and the expansion of the steel ball depends on the rotation of the bottom friction disc cycle and the intermittent translation to make the steel ball The eccentric motion achieves the effect of unfolding the surface of the steel ball to be inspected. However, the acquisition of images by this method is random and does not guarantee complete coverage of the surface of the steel ball. The mechanism is also complicated. Summary of the invention

The technical problem to be solved by the present invention is to provide a problem that can solve the problem of high speed and precise deployment of the steel ball surface. A method and apparatus for unfolding a steel ball surface based on a multi-image sensor that enhances the reliability of the overall coverage of the ball surface.

The technical solution adopted by the invention is: A method and a device for unfolding a steel ball surface based on a multi-image sensor. Based on the multi-image sensor-based steel ball surface unfolding method, a linear guide rail formed by a groove is provided, and the linear guide rail has a slope for causing the steel ball to be measured to perform a pure rolling motion along the linear guide rail by gravity, in the linear guide rail. Two image sensors of the same symmetry and the same structure are respectively disposed on the obliquely upper sides of the two sides of the axis, respectively adjusting the corresponding two image sensor parameters and the two lens optical parameters are identical, in order to obtain a spherical crown surface with a symmetrical contour, which will be The steel ball is placed on a linear orbit with a groove. The steel ball under test relies on gravity to make a one-dimensional motion along a linear track with grooves. When the steel ball under test passes under the image sensor, the two image sensors are realized. The steel ball to be tested is continuously taken.

The groove is a V-shaped groove.

The angle between the lens of the image sensor and the vertical direction of the two image sensors is ^ the angle 0 satisfies the following conditions:

Where r is the radius of the spherical surface of the image taken by the two image sensors; it is the radius of the steel ball, and A is the proportional coefficient k = r I R .

The invention relates to a device for unfolding a steel ball surface based on a multi-image sensor, comprising a worktable, a linear guide rail formed by a groove disposed in a middle portion of the worktable, and symmetrically disposed at both ends of the worktable and located on both sides of the linear guide rail The two guide rails have a slope for causing the steel ball to be tested to perform a pure rolling motion along the linear guide rail by gravity, and the top ends of the two brackets are respectively provided with an image sensor, and the two The lenses of the two image sensors on the bracket correspond to the central axis of the bottom of the linear guide.

The groove forming the linear guide is a V-shaped groove.

The angle between the lens of the camera of any one of the two image sensors and the vertical direction is ^ The angle 0 meets the following conditions:

among them,

r is the radius of the spherical surface of the image taken by the two image sensors; it is the radius of the steel ball, and A is the proportional coefficient A = WR. The multi-image sensor based steel ball surface unfolding method and device of the invention adopts a multi-image sensor based visual detection technology to propose a method for realizing the full coverage of the steel ball surface based on the multi-image sensor visual inspection technology. The image of the moving steel ball is designed to cover the surface of the steel ball at different times. At the same time, the mechanical structure of the steel ball is greatly simplified, that is, the two-dimensional relative motion is simplified to a mechanical motion with only one-dimensional pure rolling, avoiding The use of a mechanically complex, easy-to-wear, under-expanded deployment wheel system overcomes the possibility of slippage in the movement of the steel ball, solves the problem of high-speed and precise deployment of the steel ball surface, and enables the effective application of vision-based digital image processing technology. An automatic detection system for the surface quality of steel balls, improving the reliability of the automatic detection system, improving the reliability of the steel ball unfolding system, and enhancing the overall surface of the ball The reliability of the coverage is such that it no longer becomes a random probability event. The mechanical device of the invention is simpler, greatly reduces the difficulty and cost of automatic detection of surface defects of the steel ball, and has convenient operation and wide application prospect. DRAWINGS

Figure 1 is a perspective view showing the overall structure of the apparatus of the present invention;

Figure 2 is a schematic view of the two-dimensional structure of Figure 1;

Figure 3 is a perspective view of the full coverage calculation of the camera lens;

Figure 4 is a schematic view of the spherical crown surface development track;

Figure 5 is a schematic diagram of the full coverage of the vertical section of the sphere.

among them:

1: camera 2 : lens

3: Bracket 4: Workbench

5: Linear guide 6: Steel ball to be tested. BEST MODE FOR CARRYING OUT THE INVENTION The method and apparatus for unfolding a steel ball surface based on a multi-image sensor of the present invention will be described in detail below with reference to the embodiments and the accompanying drawings.

The multi-image sensor based steel ball surface unfolding method of the present invention is to provide a linear guide formed by a groove, and the groove is a V-shaped groove. The linear guide rail has a slope for causing the steel ball to be tested to perform a pure rolling motion along the linear guide rail by gravity, and two image sensors of the same name and the same structure are respectively disposed on the obliquely upper sides of the axis of the linear guide rail. The corresponding two image sensor parameters and the two lens optical parameters are respectively adjusted to obtain a spherical crown surface with a symmetric equal width, and a light source is disposed above the linear guide rail. The steel ball to be tested is placed on a linear orbit with a groove, and the steel ball to be tested is subjected to one-dimensional movement along a linear track having a groove by gravity, and two image sensors are used when the steel ball to be tested passes under the image sensor. Achieve continuous mining of the steel ball under test.

The image sensor of the present invention includes a CCD image sensor, a CMOS image sensor, and the like. The image sensor used in this embodiment is an area array CCD camera.

The angle between the lens of the CCD camera and the vertical direction of the two CCD cameras is 0, and the angle 0 satisfies the following conditions:

Where r is the radius of the spherical surface of the image taken by two CCD cameras; it is the radius of the steel ball, and A is the proportional coefficient k = r IR.

The two CCD cameras are numbered as K, K. The image of the two spherical crowns is obtained by two CCD cameras. As shown in Fig. 3, the spherical surface '' is the spherical crown surface collected by J, and ^' is 1 (the collected spherical crown surface. There is a part of the overlapping area, there is also a Part of the area below the hemisphere, the measured steel ball makes a one-dimensional pure rolling motion along the linear orbit. When the shooting areas of the two cameras coincide at the same time and both exceed the hemisphere, the surface of the steel ball can be collected without any omission. In this way, the surface of the steel ball can be fully deployed. By adjusting the proper overlap area and beyond the hemisphere area, an optimal unfolding state can be obtained, that is, while the camera is fully deployed, the number of camera acquisitions is minimized. Figure 4 is a two-dimensional view of the unfolded trajectory of the spherical surface of the shot taken over time.

As shown in Fig. 3, the spherical surface '' is the spherical crown surface collected by J, ^' is 1 (the spherical crown surface is collected. The two spherical crown surfaces intersect at O and O'. Let the two spherical crowns overlap. S<', facet ^404, the center of the circle is marked as Ν, the center of the circle is marked as Μ, the midpoint of 00' is denoted by Ρ, and the angle between the connection 0'N and 00' is ". According to the three-dimensional geometry, 0 can be launched. The smaller the smaller the coincidence area is, the larger the proportional coefficient A is, the larger the coincidence area is. The coincident part is the sufficient condition for each dead zone without the dead zone. The two overlapping circular sections can be used to quantitatively analyze two The coincidence of the circular cuts. The title is available in the MV"

In,

According to the geometric formula in A NO'

Sin « = tan * · R ² -r ² / r = tan (

I k (3) Expand the two overlapping three-dimensional spherical sections into a two-dimensional plan as shown in Figure 3. Then the critical position of coincidence is = τ/2, and the maximum position of coincidence is =0. that

From the unfolding trajectory, to achieve full coverage, there is no dead zone between different moments of acquisition, and it must be based on the narrowest point of the single acquisition sphere, ie 00'. Only a circle along the 00' sphere is crossed, and at the same time, the covered sphere beyond the hemisphere will ensure that the entire sphere is fully covered. As can be seen from Figure 4

00' = 2r cos = 2R- -sin ² 6>/cos θ (5) The spherical surface covered is in the critical condition of the hemisphere (shown in Figure 5) _C0S = r/2R, so the range of 0 beyond the hemisphere is required

Arccos k < θ < π / 2 (6) In summary, simultaneous (4), (5), (6), have

Where n is the number of times to be collected when the full coverage is expanded.

To solve this system of equations, only when it is 0.71, will the equation be satisfied. The larger the value of A is, the larger the selectable 0 range is. When 0 is fixed, the number of acquisitions is smaller as the value of A increases. Therefore, the A value that satisfies the condition is fixed, and there are different overlapping areas with the 0 position. When the coincident part is the largest, the required number of acquisitions is the smallest. The most efficient. In order to establish an intuitive concept, the table below gives the number of acquisitions required for some specific values. Table 1 Part of the parameters required to collect the number of times

In summary, adjust the parameters of the camera and lens to fix a threshold. Choose the best 0 value within the selectable range, and finally get the minimum number of full coverage acquisitions. At the same time, it is also proved that in the time of one-dimensional movement of the steel ball under the track, multiple imaging methods can be used to fully expand the surface of the sphere without dead zones. The complexity of the mechanical unfolding mechanism is greatly reduced, and the one-dimensional relative motion replaces the two-dimensional mechanical motion.

As shown in FIG. 1 and FIG. 2, the apparatus for the surface expansion method of a steel ball based on a multi-image sensor of the present invention comprises a table 4, a linear guide 5 formed by a groove provided in the middle of the table 4, The groove forming the linear guide 5 is a V-shaped groove, and two brackets 3 symmetrically disposed at both ends of the table 4 and located on both sides of the linear guide 5, the linear guide 5 having the steel ball to be tested 6 by gravity along the linear guide 5 forward slope of the pure rolling motion, the top ends of the two brackets 3 are respectively provided with an image sensor 1, the lens 2 of the two image sensors 1 on the two brackets 3 Both correspond to the central axis of the bottom of the linear guide 5. A light source 设置 is disposed above the linear guide 5 .

The angle between the lens 2 of the image sensor 1 and the vertical direction of the image sensor 1 is an angle of 0, which satisfies the following conditions:

"

"I arc co sk < θ < π I 2 Where r is the radius of the spherical surface of the image taken by the two image sensors; it is the radius of the steel ball, and A is the proportional coefficient k = r IR.

The apparatus for the surface unfolding method of the steel ball based on the multi-image sensor of the present invention, when testing the surface of the steel ball, placing the steel ball to be measured at the high end of the linear guide 5, and the steel ball to be tested depends on the linear guide 5 by gravity The front (low end) makes a pure rolling motion, so that two CCD cameras can continuously capture the steel ball under test.

Claims

Rights request

A method for unfolding a steel ball surface based on a multi-image sensor, characterized in that a linear guide formed by a groove is provided, the linear guide having a steel ball to be measured to be pure along the linear guide by gravity The slope of the rolling motion is respectively provided with two image sensors of the same name and the same structure on the obliquely upper sides of the axis of the linear guide rail, respectively adjusting the corresponding two image sensor parameters and the two lens optical parameters are consistent, in order to obtain symmetry, etc. The large spherical crown, the steel ball to be measured is placed on a linear orbit with a groove, and the steel ball to be measured relies on gravity to make a one-dimensional motion along a linear orbit with a groove, when the steel ball to be tested passes under the image sensor At the same time, the two image sensors realize continuous image capturing of the steel ball to be tested.

2. The method according to claim 1, wherein the groove is a V-shaped groove.

The multi-image sensor-based steel ball surface unfolding method according to claim 1, wherein the angle between the lens of the image sensor and the vertical direction of the two image sensors is The angle 0 satisfies the following conditions:

A device for a method for unfolding a steel ball surface based on a multi-image sensor according to claim 1, characterized by comprising a table (4) and a groove formed in a middle portion of the table (4) The linear guide (5) and the two brackets (3) symmetrically disposed at both ends of the table (4) and located on both sides of the linear guide (5), the linear guide (5) having the steel ball to be tested ( 6) relying on gravity to make a pure rolling motion along the linear guide (5), and an image sensor (1) is disposed at the top of the two brackets (3), and the two brackets (3) are The lenses (2) of the two image sensors (1) correspond to the central axis of the bottom of the linear guide (5).

The apparatus for surface expansion method of a steel ball based on a multi-image sensor according to claim 4, wherein the groove forming the linear guide (5) is a V-shaped groove.

The apparatus for a multi-image sensor based steel ball surface unfolding method according to claim 4, wherein the lens (2) and the vertical of the camera of any one of the two image sensors (1) The angle in the straight direction is ^ The angle 0 meets the following conditions:

r is the radius of the spherical surface of the image taken by the two image sensors; it is the radius of the steel ball, and A is the proportional coefficient A = WR.