-
The invention relates to a method and apparatus for processing
video pictures, especially for false contour effect compensation.
More specifically the invention is closely related to a kind
of video processing for improving the picture quality of pictures
which are displayed on matrix displays like plasma display
panels (PDP) or display devices with digital micro mirror
arrays (DMD).
Background
-
Although plasma display panels are known for many years,
plasma displays are encountering a growing interest from TV
manufacturers. Indeed, this technology now makes it possible
to achieve flat color panels of large size and with limited
depths without any viewing angle constraints. The size of the
displays may be much larger than the classical CRT picture
tubes would have ever been allowed.
-
Referring to the latest generation of European TV sets, a lot
of work has been made to improve its picture quality. Consequently,
there is a strong demand, that a TV set built in a
new technology like the plasma display technology has to provide
a picture so good or better than the old standard TV
technology. On one hand, the plasma display technology gives
the possibility of nearly unlimited screen size, also of attractive
thickness, but on the other hand, it generates new
kinds of artefacts which could damage the picture quality.
Most of these artefacts are different from the known artefacts
occurring on classical CRT color picture tubes. Already
due to this different appearance of the artefacts makes them
more visible to the viewer since the viewer is used to see
the well-known old TV artefacts.
-
The invention deals with a specific new artefact, which is
called "dynamic false contour effect" since it corresponds to
disturbances of gray levels and colors in the form of an apparition
of colored edges in the picture when an observation
point on the matrix screen moves. This kind of artefact is
enhanced when the image has a smooth gradation like when the
skin of a person is being displayed (e. g. displaying of a
face or an arm, etc.). In addition, the same problem occurs
on static images when observers are shaking their heads and
that leads to the conclusion that such a failure depends on
the human visual perception and happens on the retina of the
eye.
-
Two approaches have been discussed to compensate for the
false contour effect. As the false contour effect is directly
related to the sub-field organization of the used plasma
technology one approach is to make an optimization of the
sub-field organization of the plasma display panels. The sub-field
organization will be explained in greater detail below
but for the moment it should be noted that it is a kind of
decomposition of the 8-bit gray level in 8 or more lighting
sub-periods. An optimization of such a picture encoding will
have, indeed, a positive effect on the false contour effect.
Nevertheless, such a solution can only slightly reduce the
false contour effect amplitude but in any cases the effect
will still occur and will be perceivable. Furthermore, sub-field
organization is not a simple matter of design choice.
The more sub-fields are allowed the more complicated will the
plasma display panel be. So, optimization of the sub-field
organization is only possible in a narrow range and will not
eliminate this effect alone.
-
The second approach for the solution of above-mentioned problem
is known under the expression "pulse equalization technique".
This technique is described e.g. in Euro Display
1996, "An Equalising Pulse Technique for Improving the Gray
Scale Capability of Plasma Displays", K. Toda et al., pages
39 to 42. This technique is a more complex one. It utilizes
equalizing pulses which are added or separated from the TV
signal when disturbances of gray scales are foreseen. In addition
for better compensation quality, since the fact that
the false contour effect is motion relevant, different pulses
for each possible speed are needed. That leads to the need of
a big memory storing a number of big look-up tables (LUT) for
each speed and there is a need of a motion estimator. A problem
with these equalizing pulses is that they are used to increase
or decrease the amplitude of the video signal in areas
where false contour effect is likely to occur. Thus the correction
value is added to the pixel value (RGB data for
Plasma Displays) before the corresponding sub-field code word
is calculated. Therefore, its not taken into account at which
position within the frame period a sub-field is inserted or
omitted.
Invention
-
Therefore, it is an object of the present invention to disclose
a method and an apparatus which is based on the known
solutions using equalizing pulses but which allows for a more
efficient false contour effect compensation. This object is
achieved by the measures claimed in claims 1 and 4.
-
The general idea of the invention is that the correction of
pixel values is made not on amplitude values only without
consideration of the position of the sub-fields which are inserted
or omitted but on sub-field level. When the motion in
the picture is known for the pixels then the sub-fields for
correction are positioned at the best possible location in
the frame period for false contour effect compensation.
-
A correction performed on subfield level allows directly to
insert or to remove subfields on the position (time position
within the frame) where too much or not enough light impulses
are available. This way it's possible to compensate directly
the failures where they occur.
-
Advantageously, additional embodiments of the inventive
method are disclosed in the respective dependent claims.
-
One example for an apparatus according to the invention is
disclosed in claim 3. With a motion estimator the apparatus
calculates motion vectors for blocks of pixels of the video
frames. It also comprises means for determining critical
pixel value transitions which are moving. For given motion
vectors and critical pixel value transitions look-up tables
are provided in which the corrected digital code words are
stored which are to be used for a good false contour effect
compensation.
Drawings
-
Exemplary embodiments of the invention are illustrated in the
drawings and are explained in more detail in the following
description.
-
In the figures:
- Fig. 1
- shows a video picture in which the false contour
effect is simulated;
- Fig. 2
- shows an illustration for explaining the sub-field
organization of a PDP;
- Fig. 3
- shows an illustration for explaining the false contour
effect;
- Fig. 4
- illustrates the appearance of a dark edge when a
display of two frames is being made in the manner
shown in Fig. 3;
- Fig. 5
- shows two different sub-field organization schemes;
- Fig. 6
- shows the illustration of Fig. 3 but with sub-field
organization according to Fig. 5;
- Fig. 7
- shows the effect on eye retina for the amplitude
based correction of the false contour effect;
- Fig. 8
- shows the effect on the eye retina for the amplitude
based correction illustrated with sub-field
resolution;
- Fig. 9
- shows the video picture of Fig. 1 with a subdivision
in blocks of pixels;
- Fig. 10
- shows the effect on the eye retina for the sub-field
based correction method illustrated with sub-field
resolution;
- Fig. 11
- shows a block diagram of the apparatus according to
the invention.
Exemplary embodiments
-
The artefact due to the false contour effect is shown in Fig.
1. On the arm of the displayed woman are shown two dark
lines, which e. g. are caused by this false contour effect.
Also in the face of the woman such dark lines occur on the
right side.
-
A plasma display panel utilizes a matrix array of discharge
cells which could only be switched ON or OFF. Also unlike a
CRT or LCD in which gray levels are expressed by analog control
of the light emission, in a PDP the gray level is controlled
by modulating the number of light pulses per frame.
This time-modulation will be integrated by the eye over a period
corresponding to the eye time response. When an observation
point (eye focus area) on the PDP screen moves, the eye
will follow this movement. Consequently, it will no more integrate
the light from the same cell over a frame period
(static integration) but it will integrate information coming
from different cells located on the movement trajectory. Thus
it will mix all the light pulses during this movement which
leads to a faulty signal information. This effect will now be
explained in more detail below.
-
In the field of video processing is an 8-bit representation
of e.g. a luminance level very common. In this case each
level will be represented by a combination of the following 8
bits:
20 = 1, 21 = 2, 22 = 4, 23 = 8, 24 = 16, 25 = 32, 26 = 64,
27 = 128
-
To realize such a coding scheme with the PDP technology, the
frame period will be divided in 8 lighting periods which are
also very often referred to sub-fields, each one corresponding
to one of the 8 bits. The number of light pulses for the
bit 21 = 2 is the double of that for the bit 20 = 1, etc..
With a combination of these 8 sub-periods, we are able to
build said 256 different gray levels. Without motion, the eye
of the observer will integrate over about a frame period
these sub-periods and will have the impression of the right
gray level. The above-mentioned sub-field organization is
shown in Fig. 2.
-
The light emission pattern according to the sub-field organization
introduces new categories of image quality degradation
corresponding to disturbances of gray levels and colors. As
already explained, these disturbances are defined as so-called
dynamic false contour effect since the fact that it
corresponds to the appearance of colored edges in the picture
when an observation point on the PDP screen moves. The observer
has the impression of a strong contour appearing on a
homogeneous area like displayed skin. The degradation is enhanced
when the image has a smooth gradation and also when
the light emission period exceeds several milliseconds. So,
in dark scenes the effect is not so disturbing as in scenes
with average gray level (e.g. luminance values from 32 to
223).
-
In addition, the same problem occurs in static images when
observers are shaking the heads which leads to the conclusion
that such a failure depends on the human visual perception.
-
To better understand the basic mechanism of visual perception
of moving images, a simple case will be considered. Let us
assume a transition between the luminance levels 128 and 127
moving at a speed of 5 pixel per video frame and the eye is
following this movement. Fig. 3 shows a darker shaded area
corresponding to the luminance level 128 and a lighter shaded
area corresponding to the luminance area level 127. The sub-field
organization, shown in Fig. 2 is used for building the
luminance levels 128 and 127 as it is depicted on the right
side of Fig. 3. The three parallel lines in Fig. 3 indicate
the direction in which the eye is following the movement. The
two outer lines show the area borders where a faulty signal
will be perceived. Between them the eye will perceive a lack
of luminance which leads to the appearance of a dark edge in
the corresponding area which is illustrated in Fig. 4. The
effect that a lack of luminance will be perceived in the
shown area is due to the fact that the eye will no more integrate
all lighting periods of one pixel when the point from
which the eye receives light is in movement. Only part of the
light pulses will probably be integrated when the point
moves. Therefore, there is a lack of corresponding luminance
and the dark edge will occur. On the left side of Fig. 4,
there is shown a curve which illustrates the behavior of the
eye cells during observing the moving picture depicted in
Fig. 3. The eye cells having a good distance from the horizontal
transition will integrate enough light from the corresponding
pixels. Only the eye cells which are near the transition
will not be able to integrate a lot of light from the
same pixels.
-
To improve this behavior at first, a new sub-field organization
is presented which has more sub-fields and above all has
more sub-fields with the same weight. This will already reduce
the contouring effect and improve the situation. Furthermore,
it allows for the inventive correction method which
will be explained afterwards. In Fig. 5 two examples of new
coding schemes are shown. The choice of the optimal one has
to be made depending on the plasma technology. In the first
example there are ten sub-fields used wherein there are four
sub-fields having lighting periods with a relative duration
of 48/256. In the second example there are twelve sub-fields
and there are seven sub-fields having the relative duration
of 32/256. Please note that the frame period has a relative
duration of 256/256.
-
In Fig. 6 the result of the new sub-field organization according
to the second example of Fig. 5 is shown in case of
the 128/127 horizontal transition moving at a speed of five
pixels per frame. Now, the chance that the corresponding eye
cells will integrate more similar amounts of lighting periods
is increased. This is illustrated by the eye-stimuli integration
curve at the bottom of Fig. 6 when compared to the eye-stimuli
integration curve at the bottom of Fig. 3.
-
For false contour reduction some solutions exist in which
correction signals are added to the video signal in order to
compensate the lack of luminance (dark edges) or the increase
of luminance (luminous edges) . All the solutions known reduce
or increase the amplitude of the video signal in areas where
false contour occurs.
-
The following example explains the used principle:
It is assumed that an 3 x 8 bit coded RGB picture is converted
to 12 bit sub-field codes. This conversion is realized
for example by a LUT (Look Up Table) in which the 12 bit sub-field
codes are stored for the different 8 bit RGB data
words. In this way the the video signal (3 times for RGB) is
converted into the sub-field code of 12 bit for each color
channel.
-
The known false contour correction methods (with equalizing
pulses) correct directly the pixel values of the video signal,
i.e. correction is done before the sub-field conversion.
-
An illustration of this method is shown in Fig. 7. From Fig.
7a) it follows that in the middle of the transition the amplitude
on the eye retina has a lack of 32 relative amplitude
units. This is compensated by simply adding this value to the
pixels of the transition, see Fig. 7b). Since the brightness
impression on the eye is given by the integration of the
light amplitude over a certain time period, such a correction
cannot be perfect when the eye moves.
-
The effect on sub-field level after generation of the sub-field
code words is shown in Fig. 8. For three pixels of the
transition an additional sub-field with weight 32 corresponding
to the correction value +32 is activated (see the dark
black bars shown in Fig. 8). Note, that only three pixels of
the transition have the additional sub-field of weight 32.
This is because the transition would otherwise be distorted.
-
The eye stimuli integration curve shown at the bottom of Fig.
8 indicates that the false contour effect is reduced compared
to Fig. 6 but still present.
-
The disadvantage of the amplitude correction can also be seen
on the table below. Taking the previous example, a correction
value of 32 can have an influence on different timing positions,
e.g. SF 9 or SF10.
-
The effect of the two corrections shown in the table implemented
(add of the sub-field No. 9 or 10 with both a value of
32) are totally different for the eye and consequently for
the impression of picture brightness but they both have the
same amplitude of 159.
Sub-Field | SF0 | SF1 | SF2 | SF3 | SF4 | SF5 | Ampl. |
Corr. 1 | 1 | 2 | 4 | 8 | 16 | 32 |
Corr. 2 | 1 | 2 | 4 | 8 | 16 | 32 |
Sub-Field | SF6 | SF7 | SF8 | SF9 | SF10 | SF11 | Ampl. |
Corr. 1 | 32 | 32 | 32 | 0 | 0 | 0 | 159 |
Corr. 2 | 32 | 32 | 0 | 32 | 0 | 0 | 159 |
-
Already for the compensation technique used for in Fig. 8 its
necessary to have knowledge about the movement in the picture
and where the critical transition is located. A motion estimator
is applied for providing motion vectors of blocks of
pixels. At first, the original picture is segmented in
blocks, each of which will have a single motion vector assigned.
An example of such a decomposition is shown in Fig.
9. Other types of motion-dependent pictures segmentations
could be used, since the goal is only to decompose the picture
in basic elements having a well-defined motion vector.
So all motion estimators can be used for the invention, which
are able to subdivide a picture in blocks and to calculate
for each block a corresponding motion vector. As motion estimators
are well-known from, for example 100 Hz up-conversion
technique and also from MPEG coding etc., they are well-known
in the art and there is no need to describe them in greater
detail here. As an example where a motion estimator is described
which could be used in this invention, it is referred
to WO-A-89/08891. Best to be used are motion estimators which
give precisely the direction of the movement and the amplitude
of this movement for each block. Since most of the
plasma display panels are working on RGB component data,
benefit could be achieved when for each RGB component a separate
motion estimation is being carried out and these three
components are combined so that the efficiency of the motion
estimation will be improved. In another block it is evaluated
whether two adjacent blocks have the same motion vector in
order to find the critical pixel value transitions which
could cause false contours. Additionally each block can be
evaluated for critical transitions. A critical transition is
found when two areas of pixels with slightly different pixel
values are found. Here, most of the sub-fields of the two
pixel value code words are identical except for one sub-field
with greater weight and a number of sub-fields with smaller
weight (see e.g. Fig. 6).
-
A correction performed on sub-field level according to the
invention allows directly to insert or to remove subfields on
the position (time position within the frame) where too much
or not enough light impulses are available. This way it's
possible to compensate directly the failures where they occur.
-
In case of a sub-field based compensation, subfields are inserted
or removed depending on the transition and the speed
of movement. That means that it's directly possible to insert
or remove light pulses on positions (in temporal direction)
where they are missing or are too much. The main difference
to the amplitude based compensation is that with the amplitude
based compensation technique it is not possible to determine
the time where the additional light pulses are best
to be inserted or removed.
-
In Fig. 10, the subfield-based compensation technique is depicted
with an example. The additional subfields are shown
with small black boxes. For the first pixel having a pixel
value of 127 the sub-field with weight 16 is omitted, also
for compensation reason. The correction depicted in Fig. 10
is an example for a good false contour effect compensation
for this transition and movement. The additional subfields
are shown with small black boxes, generate light pulses exactly
in the time period where they are needed. Within the
area of the parallel lines shown, the eye will perceive light
emission pulses of total weight ≈128 when looking along the
shown direction. But it is to be noted that the integration
of the eye retina is also a function of time distance between
the sub-fields. So an easy way to find the best results for
compensation fo a given transition with a given motion vector
is to make experiments.
The video processing block used to compensate the false contour
effect is shown in Fig. 11. Reference number 10 denotes
the whole block. RGB data is input to this block. After initilisation
one frame N will be stored in frame memory 11 and
data of frame N+1 will be delievered to a motion estimation
and transition detection unit 12. Within this unit the picture
is subdivided in blocks and motion vectors are calculated
for the blocks. Preferably the subdivision in blocks is
made so that all pixels in the blocks have identical pixel
values. When the motion vectors are found, critical transitions
are searched. This can be done by looking for adjacent
blocks with identical motion vectors and pixel values to
which sub-field codes correspond which have a difference
mainly in sub-fields of greater weight, see above given explanation.
Also the found transition will be classified with
regard to the pixel value differences of the transition.
-
The information regarding the motion vector and the transition
classification is fed to look up table memory 13. In
look up table memory 13 a number of look up tables 14 are
stored. The information regarding motion vector and transition
classification serves as an address for the right table.
From the information found during transition detection a control
signal is generated which controls which entry in the
selected look up table is to be output. For the pixels of the
transition which are to be corrected new sub-field codes are
stored in the look up table and these codes are read out under
control of this signal. Another control signal is generated
for the control of a demultiplexer 15 at the output of
the look up table. This signal is used to switch between the
output of the look up table 14 and the output of sub-field
code generation unit 16 in which the RGB pixel values of a
frame are converted to sub-field codes. Another look up table
can be used for this purpose. As a result, at the output of
look up table unit 13 the sub-field codes of the frame are
supplied to the display unit inclusive the corrected sub-field
codes for the critical moving transitions.
-
The invention is not restricted to the disclosed embodiments.
Various modifications are possible and are considered to fall
within the scope of the claims. E.g. a different sub-field
organization could be used. The values in implementations
covered by the patent may differ from those here shown, in
particular the number and weight of the used sub-fields.
-
An alternative embodiment is one without motion estimator.
Here, the pixel values of two succeeding frames are compared
pixel by pixel and each time, a critical difference is found
a corresponding corrected sub-field code is selected in a
look up table. With this simple solution the correction results
will be less good as in the example explained above,
but for a low cost implementation the solution may be sufficient.
-
All kinds of displays which are controlled by using different
numbers of pulses for gray-level control can be used in connection
with this invention.