JPH10333638A - Dynamic image display method and dynamic image display device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To suppress the occurrence of a false contour when a dynamic image is followed by eyes by correcting the present image data in response to the movement of picture elements near the signal level generated by the false contour. SOLUTION: A moving vector detection section 25 is constituted of a binary arithmetic block 26 provided with binary arithmetic sections for individual colors and a comparing/detecting block 27 provided with comparison sections for individual colors, and the movement of an image between fields is obtained by block matching. A majority decision integrated judgment section 28 judges the number of moving picture elements and moving direction of each detecting block KB and registers the judged result in a moving vector table 29. A data correction process section 4 corrects the picture element data of a false contour section with a correction table generated based on visual experiments in advance. The movement of the picture elements near the signal level generated by the false contour is grasped, and the present image data are corrected in response to the movement of the picture elements.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a moving image display method and apparatus for suppressing the occurrence of false contours that occur when an image moves in a plasma display panel (hereinafter simply referred to as "PDP") or the like. .

[0002]

2. Description of the Related Art In response to recent demands for larger display devices, thin matrix panels such as PDPs, EL display devices, fluorescent display tubes, and liquid crystal display devices have begun to be provided. Among such thin display devices, PDP is particularly expected to be a large-screen, direct-view display device.

Incidentally, there is an intra-field time division method as one of the halftone display methods of the PDP. In this halftone display method, one field is composed of N screens having different luminance weights (these screens are hereinafter referred to as subfields). SF0, SF1, SF2,...
, SF (N-1), and the ratios of the luminance weights of the subfields are 20, 20, 22, 22,..., Respectively.
2N-1. Intermediate luminance in one field is determined by selecting the presence or absence of light emission in these subfields. The luminance perceived by the human eye is determined by the human visual characteristics (afterglow characteristics). It can be expressed as a sum. The number of halftones that can be expressed by this halftone display method is the number of subfields in one field, that is, 2N.

FIG. 21 shows a display sequence in one field using the halftone display method. 1 field is 8
(N = 8) subfields having different luminance weights, and SF7, SF6, SF6,
..., SF0. Here, SF7 is the most significant bit (MSB) side, and SF0 is the least significant bit (LSB).
Called B) side. Each sub-field is often used to control light emission in one field in the order of SF0, SF1,... That is, assuming that the ratio of the number of times of light emission in each subfield is "1" for SF0, SF1 is "2",
SF2 is "4",... SF6 is "64", and SF7 is "128". When the number of subfields is 8, up to 256 gradations can be expressed.

The halftone display method based on the above-described subfield method is an excellent technique as a technique capable of expressing multiple gradations even in a binary display device such as a PDP which can express only two gradations of 1 and 0. With the display method using the subfields, almost the same image quality as a television image can be obtained in a PDP.

However, for example, when a moving image is displayed on a subject whose gradation is gradually changing,
There is a problem that a so-called false contour peculiar to a PDP image which cannot be seen on a television occurs. This false contour generation phenomenon is caused by the visual characteristics. When the video signal level is displayed in 256 gradations, it is as if the floor is along the 2Nth boundary such as 128, 64, 32, 16 or the like. This is a phenomenon in which a color different from the color to be originally displayed appears as stripes in a state where the tone is lost. However, when a still image is displayed, no false contour is perceived. It is a feature of the false contour that it is perceived only in the moving part and around the signal level.

Referring to FIG. 22, the principle of generating false contours in the gradation display method by the subfield method will be described.
In FIG. 22A, the number of subfields in one field is eight, and the arrangement thereof is smaller in luminance weight, that is, SF0, SF0,
.., SF7 are arranged in this order. When the signal level at a certain pixel position changes from 127 to 128, the moving image moves by three pixels in one field. FIG. 22B shows the result of the observer observing this moving image on the screen.

As described above, when the signal level 127 (light emission from SF0 to SF6) and the signal level 128 (only SF7 emits light) are adjacent to each other, the gradation difference is 1 LSB.
However, the light emission value perceived on the human retina is equal to (1/256), but as a result of the non-uniformity of the light emission time, the light emission of each signal level overlaps by the amount of the moved pixel of the image, resulting in spatial expansion and retina. It is felt as a large value (integral value) above. In other words, the light emission of each subfield, which should originally emit light at the same pixel, is emitted at a different pixel position in the moving image part, and the halftone luminance of the pixel can be expressed simply by the sum of each subfield. Disappears. This is why it is felt as a false contour.

As shown in FIG. 22, when the moving image is scrolled from the left side to the right side of the display screen, the above-described boundary of the signal level is felt as a bright line, and conversely, the moving image moves from the right side to the left side of the display screen. When scrolling, the signal level boundaries described above will be perceived as spatial separation of subfields and will be perceived as dark lines. On the other hand, the arrangement of the subfields has the larger luminance weight, that is, SF
7, SF6, SF5,..., SF0, when the moving image is scrolled from the left side to the right side of the display screen, the boundary of the signal level is perceived as a dark line, and conversely, the moving image is displayed. When scrolling from the right side of the screen to the left side, the boundaries of the signal level will be felt as bright lines. That is, the appearance of the false contour differs depending on the moving direction of the moving image on the display screen.

Further, the generation of the false contour also depends on the moving speed of the moving image, and the higher the moving speed, the larger the range over which the false contour reaches. For example, in a moving image that moves by 10 pixels in one field, the pixel width covered by the false contour is as large as 10 pixels.

Conventionally, various proposals have been made as a countermeasure against this false contour. In Japanese Patent Application Laid-Open No. Hei 7-271325, the display order of subfields is determined by setting the pulse number ratio to 1, 2, 4, 8, 16, There is disclosed a technique of rearranging in order such that false contours are not conspicuous, instead of simple increases such as 32, 64, and 128. For example, there is a method of displaying the subfield having the longest display period among the subfields in an order such that the subfield is arranged at the center of the field, or changing the display order for each field.

However, if the subfield rearrangement or the subfield light emission sequence is changed for each field, the effect is extremely limited, such as the inability to cope with a false contour of a fast-moving moving image.

Japanese Patent Application Laid-Open No. 8-123355 discloses a technique for suppressing false contours using motion detection. In this method, a motion amount and a direction are obtained from a screen corresponding to a moving image and a background image of two continuous field screens, and a motion correction amount is obtained based on the detected value and a division time ratio in a unit time of each subfield screen. In this technique, the light emission pattern of the subfield screen corresponding to the correction amount is moved.

In Japanese Patent Application Laid-Open No. Hei 8-212848, a motion vector is detected for each pixel block by display data between fields, and a head subfield in the field displays data corresponding to the input data, and is followed by the data. For each subfield, a technique is disclosed in which an image is displayed by moving display data by using a value obtained by multiplying a motion vector by a value obtained by dividing a delay time from each head subfield by a field period and a motion vector.

By simply moving the light emission pattern of the subfield or changing the display data in accordance with the amount of movement as described above, unnatural gradation changes and whiskers occur at the block boundaries of the moving image, and adverse side effects are conspicuous. There are cases. Further, it was found from a visual experiment that matching with the visual light amount was also difficult, and it was not possible to cope with the movement amount only by moving the data. Also, in false contour suppression using motion detection, a practical configuration of motion detection with practicality is irrespective of how accurately a motion amount is detected in order to prevent false contours. Are not fully disclosed.

In the false contour correction method disclosed in Japanese Patent Application Laid-Open No. H8-234694, the unit pixel data corresponding to the same pixel is compared with the previous value separated by at least one frame period and the current value. When the digit position of the most significant bit of the light emission logical value of the above is different from each other, the correction data is added to or subtracted from the current value.

However, in this false contour correction method, the effect may be reversed if the moving direction of the moving image cannot be specified. For example, when the digit position of the bit is detected in the upper direction, the correction data is subtracted. However, as described above, when the above operation is performed when the image is moving to the left, the false contour is emphasized and the effect is reversed. A case arises. Similarly, when a bit is detected in the lower direction, correction data is added.
The effect is reversed if the image is moving in the opposite direction.
In addition, it has a disadvantage that it cannot cope with a high-speed moving image.

[0018]

As described above, the conventional technique relating to the suppression of false contours has not been able to sufficiently cope with the prevention of false contours occurring in fast-moving moving images.

The present invention has been made in view of the above circumstances, and in a display device that performs gradation display by the subfield method, it is possible to greatly suppress the occurrence of false contours when a moving image is visually followed. It is an object to provide a moving image display method and a moving image display device.

[0020]

Means for Solving the Problems In order to achieve the above object, the present invention takes the following measures.

The first aspect of the present invention employs a configuration in which the movement of a pixel near a signal level at which a false contour occurs is captured, and the current image data is corrected in accordance with the movement of the pixel. According to this configuration, by capturing the motion of the pixel in the vicinity of the signal level where the false contour occurs, a local change in the image can be detected at high speed without being affected by the bias of the distribution of the pixel values. . Very local detection of the motion of the pixel where the false contour portion occurs is sufficient, and the calculation time, circuit configuration, and the like can be simplified.

According to a second aspect of the present invention, the current field image and the previous field image are binarized in a section near the signal level at which a false contour occurs by using a threshold value, and the binarized image is compared with a moving pixel. The number of moving pixels and the moving direction are detected, and a moving pixel having a signal level at which a false contour occurs in the current field image is corrected in accordance with the number of moving pixels and the moving direction. According to this configuration, since the current field image and the previous field image are binarized using the threshold value near the signal level at which the false contour occurs, the characteristics of the original image are reflected even in a small area. Can be detected. Further, since the correction is made in accordance with the number of moving pixels and the moving direction of the pixel whose motion has been detected, the occurrence of false contour can be prevented.

According to a third aspect of the present invention, the current field binary image is divided into a plurality of detection blocks, a reference area is set in the previous field binary image for each detection block, and a plurality of reference areas are set in the reference area. The degree of coincidence between the reference block and the detected block is evaluated, and the number of moving pixels and the moving direction of the moving pixel are detected from the positional relationship between the reference block having the highest evaluation value and the detected block. According to this configuration,
Since a motion vector of an image in which a false contour does not occur can be ignored, and a motion is detected in a binary image in the vicinity where a false contour occurs, a pixel value (multi-valued image) represented by, for example, 8 bits is used. As compared with the case, the operation time for subtraction, comparison, and the like can be significantly reduced, and it is possible to sufficiently cope with a moving image with fast movement.

According to a fourth aspect of the present invention, a detected block in which no motion is detected in a comparison between a detected block and a reference block is temporarily registered as an undefined block, and an undefined block surrounded by a known block in which a motion is detected. A configuration is adopted in which the number of moving pixels and the moving direction are the same as those of the known block. According to this configuration, the indefinite block surrounded by the known block in which the motion has been detected is treated as having the same motion as the known block, so that the plurality of blocks including the edge portion of the moving image is one or more blocks. , The motion of the enclosed block can be detected.

According to a fifth aspect of the present invention, the processing of an undefined block employs a configuration in which the number of moving pixels and the moving direction of the detection block are obtained by linear interpolation based on the number of moving pixels and the moving direction of a known block surrounding the block. According to this configuration, the undefined block surrounded by the known block in which the motion is detected is obtained by linear interpolation from the neighboring known blocks, so that a plurality of blocks including the edge portion of the moving image is replaced with one or a plurality of undefined blocks. Even in the case of surrounding, the movement of the surrounded block can be detected with higher accuracy.

According to the present invention, the number of moving pixels and the moving direction of each of the R, G, and B color components are detected for each detection block, and the number of moving pixels and the moving direction of the detected block are determined by majority decision. Take the configuration. According to this configuration, since the number of moving pixels and the moving direction of the detection block are determined by the majority decision, the motion of the block can be determined.

According to the seventh aspect of the present invention, when extracting the number of moving pixels and the moving direction of the detected block of each color component and its neighboring blocks, the color components are rearranged in descending order of numerical value, and the median value is calculated. A configuration to be obtained as a representative value is adopted. According to this configuration, the number of moving pixels and the moving direction of the detection block are determined using the number of moving pixels and the moving direction obtained for each color component and a plurality of values of the number of moving pixels and the moving direction of a neighboring block surrounding the moving pixel. Therefore, motion determination with higher accuracy can be realized.

According to the present invention, the current field image and the previous field image are binarized with a multi-level threshold centering on a signal level of 2 N or a signal level obtained by combining these signal levels. , A configuration for detecting a motion for each multi-layer image data. According to this configuration, a false contour occurs near the 2Nth signal level boundary, so that a binary image in the vicinity of each signal level at which the false contour occurs can be extracted. Is detected, false contours generated at each signal level can be suppressed.

The ninth aspect of the present invention employs a configuration in which the current field image and the previous field image are smoothed before binarization. According to this configuration, since the noise of the current field image and the previous field image can be removed, the motion can be accurately detected.

According to a tenth aspect of the present invention, the correction data corresponding to the number of moving pixels and the moving direction is registered in a correction table corresponding to the signal level at which a false contour occurs, and the number of moving pixels of the detected moving pixel is registered. And taking out correction data from the correction table based on the moving direction. According to this configuration, the number of moving pixels and the moving direction of the moving pixel can be detected, and correction can be performed using a correction table in which correction data is finely set based on the number of moving pixels and the moving direction.

According to an eleventh aspect of the present invention, the number of moving pixels and the moving direction detected for each detection block are registered in a motion vector table, and the coordinate data of a signal level at which a false contour occurs is detected from the current field image. A correction candidate pixel is extracted, and the number of moving pixels and the moving direction of the detection block to which the correction candidate pixel belongs are taken out from the motion vector table, and the number of moving pixels and the moving direction taken out correspond to the false contour occurrence signal level. The correction data is taken out from the correction table. According to this configuration, it is possible to obtain optimal correction data corresponding to the signal level, the number of motion pixels, and the direction for a pixel having a signal level at which a false contour occurs and for which motion has been detected.

According to a twelfth aspect of the present invention, a configuration is adopted in which a correction candidate pixel is determined based on the detection density of a signal level at which a false contour occurs. According to this configuration, it is possible to correct only pixels in a pixel density region where a false contour is visually confirmed.

According to a thirteenth aspect of the present invention, there is provided a binarization processing means for binarizing a current field image and a previous field image with a threshold value near a signal level at which a false contour occurs, and a binarization image between fields. And a correction table in which correction data corresponding to the number of moving pixels and the moving direction are registered for each signal level at which a false contour occurs, and a false contour of the false contour from the current field image. The present invention employs a configuration including extraction means for extracting a pixel having a generated signal level, and correction means for extracting correction data corresponding to a motion detection result and a signal level of the extracted pixel from the correction table and correcting the pixel. According to this configuration, P
A moving image display in which false contours cannot be seen in DP or the like can be reliably obtained.

[0034]

Next, embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 shows the overall configuration of a gradation image display device to which the moving image display method of the present invention is applied. In this gradation image display device, a video signal processing unit 1 separates a video signal into R, G, and B color components, and an A / D conversion unit 2 converts the video signal into R, G, and B image data. Input to the detection processing unit 3. The motion detection processing unit 3 performs smoothing processing, multi-level binarization processing, block matching processing, and majority / integration determination processing to detect the number of motion pixels and the moving direction of the input image. The number of motion pixels, the moving direction, and the detected block information detected by the motion detection processing unit 3 are input to the data correction processing unit 4 provided with a correction table in which a correction amount according to the number of motion pixels and the moving direction is set, and the data is false. The gradation data of the pixel having the contour is corrected and output to the output processing unit 5. The output processing unit 5 converts the gradation data of each pixel into a pulse number corresponding to the voltage application time width, and supplies the pulse number to the X scan driver 7 and the Y scan driver 8, thereby providing an intermediate signal to the image display unit 6 composed of a PDP. Key display is being performed.

The synchronization separation processing section 9 separates the synchronization signal from the video signal, generates a timing signal synchronized with the input video signal in the timing generation section 10 and gives the timing signal to each section. FIG. 2 shows a specific configuration of the motion detection processing unit 3.
Shown in The original image data input from the A / D converter 2 is
The signal is input to the current field block B1 and delayed by one field by the delay circuit 20 before being input to the previous field block B2. In the current field block B1, original image data of the current field is converted into a smoothing filter block 21- composed of three smoothing filters of R, G, and B.
1 and smoothing processing is performed for each color to remove noise components included in the original image. The smoothed original image data is input to a multi-level binarization block 22-1 including a multi-level binarization processing unit provided for each of R, G, and B, and a plurality of binarization thresholds are provided for each color. Performs binarization processing. Here, the binarization threshold is set in the vicinity of the signal level at which a false contour occurs. For example, when displaying 256 gradations with 8 bits, 31 to 32, 63 to 64, and 127 to 12
8 and the reverse signal level. The binarized image data (multi-layer image data) obtained for each threshold value is stored in the image memory of each color of the multi-layer memory block 23-1.

The field block B2 before processing the original image one field before has the same configuration as that of the current field block B1. That is, the original image data one field before is smoothed for each color by the smoothing filter block 21-2, and the multi-level binarization processing unit provided for each of R, G, B is provided. In block 22-2, binarization processing is performed with a plurality of binarization thresholds for each color, and the multi-layer image data obtained for each threshold is stored in the image memory of each color in the multi-layer memory block 23-2. I do.

In the block matching process for motion detection, the address mapping of the detection block KB cut out from the current field image is performed by the address mapping unit 24.
-1 is performed, and the address mapping unit 24-1 performs address mapping of the block of the reference block RB cut out from the previous field image. Each image data of the detection block KB and the reference block RB is input to the motion vector detection unit 25.

The motion vector detecting section 25 comprises a binary calculating block 26 in which a binary calculating section is provided for each color, and a comparison detecting block 27 in which a comparing section is provided for each color. Find the movement of the image between.

The majority / integration determination unit 28 determines the number of motion pixels and the direction of motion of each detection block KB, and registers the determination result in the motion vector table 29.

The data correction processing unit 4 corrects the pixel data of the false contour portion using the motion vector table 29 and a correction table created based on a visual experiment in advance.

The details of the operation of the gradation image display device configured as described above will be described. FIG. 10 is a diagram conceptually extracting the processing contents of each block of the motion detection processing unit 3 shown in FIG. Hereinafter, the processing content of each block of the motion detection processing unit 3 shown in FIG. 10 will be described.

Each of the current field image and the previous field image is subjected to a binarization process using a binarization threshold set to a signal level at which a false contour occurs.

Here, the concept of a multi-level binary image will be described with reference to FIG. FIG. 11 shows the pixel position on the horizontal axis,
The vertical axis indicates the pixel value. The domain of the pixel value is n thresholds T
From h1 to Thn (n = 5 in the figure), (n + 1)
It is divided into a plurality of sections, and binarized pixel values are assigned so that values are different in adjacent sections, and each pixel is binarized according to which section the pixel value belongs to.

Since this binarization method can represent a local change in an image without being affected by the bias of the distribution of pixel values, the binarized image can be used as a block for detecting a motion vector. Even in a small area, the characteristics of the original image are reflected. Very local detection of the motion of a pixel where a false contour occurs is sufficient, and the calculation time, circuit configuration, and the like can be simplified.

Next, the block matching processing in the motion vector detecting section 25 will be described. A block matching method is used as a method for obtaining the motion of an image between fields. This method uses the detection block K as described above.
B into 16 horizontal pixels and 16 vertical pixels, the reference block RB
Is set to the same size as the detection block KB, the reference area R including the reference block RB is 48 pixels horizontally and 4 pixels vertically.
Assuming eight pixels, the number of motion pixels in the horizontal direction (x direction) is −
From 16 pixels to +16 pixels can be detected, and similarly, the number of motion pixels in the vertical direction (y direction) can be detected from -16 pixels to +16 pixels. When the moving image moves on the display screen, the false contour is conspicuous because the movement between the fields is in the vicinity of 6 to 15 pixels. As shown in FIG. 12, the motion of the image between the current field and the previous field is obtained in units of rectangular blocks, and the best match between each detection block KB obtained by dividing the current field into a plurality of reference regions R is defined as the previous field. It is found among the reference blocks RB included in the reference region R, and the amount between them is obtained as a motion vector.

At this time, the degree of coincidence of the blocks is determined based on the magnitude of the determined evaluation function value, and the motion vector is detected from among a number of reference blocks RB included in the reference region R, which gives the minimum value of the evaluation function. We do this by finding out.

When a configuration based on the above method is performed on a grayscale image, a large number of operations such as subtraction and comparison are performed by using pixel values represented by multiple values (for example, 8 bits) in the calculation of the evaluation function value. Since the number of reference blocks RB that can be evaluated in real-time processing is limited, there is a case where a reference block RB that gives a correct motion vector is not evaluated, resulting in a problem that the detection accuracy of the motion vector is reduced. .

To suppress false contours in PDP,
Since the signal level at which the false contour occurs is known, in order to determine the motion of the image near the false contour, the image in the vicinity is binarized, and the motion vector of the binary image is determined. The calculation time and accuracy are improved. Also,
A motion vector of an image in which a false contour does not occur can be ignored.

In the binary operation block 26, an evaluation function indicating the degree of coincidence is calculated. FIG. 3 shows the configuration of the binary operation unit when the detection block KB size is 4 * 4 pixels. Note that the detection block KB size is 16 * 16, and the reference area R
Is described as 48 * 48, but here, it is described as 4 * 4 for simplicity.

The binary operation unit includes binary line matching units 30-1, 30-2, 30-3, and 30-4 for detecting matching between fields for four lines in the block. The binary line matching unit 30-1 checks the line matching of the first line between the current field and the previous field. The line data of the first line of the detection block KB is input to the shift register 31-1, and the line data of the first line of the reference block RB is input to the shift register 31-2. Shift registers 31-1, 31-
2 can hold four pieces of pixel data, and the two exclusive-OR circuits 32-1, 32-2, 32-3, and 32 have the same pixel data from the two shift registers.
-4. The output of the exclusive OR circuit is added by a bit adder circuit 33 and a multiplexer 34
Enter

Binary line matching units 30-2, 30-
3 and 30-4 are also the binary line matching unit 30-1.
, And evaluates the line matching of the second to fourth lines, and inputs a bit addition value serving as a line evaluation value to the multiplexer 34. Via the multiplexer 34, the binary line matching units 30-1, 3
0-2, 30-3 and 30-4 are selectively added to the adder 3
5 and the sum of the four line evaluation values is stored in the register 36 as a block evaluation value for one block. The register 37 is for output control.

In the above-mentioned binary operation unit, the data of the detection block KB and the reference block RB supplied from the multi-level image memory via the signal lines are sent to the binary line matching unit 30 in scanning line units. Each line matching unit 30-
In 1 to 30-4, shift registers 31-1, 31-
2 is used to extract binary data for each pixel. Pixels at the same position in the block are evaluated for coincidence and non-coincidence by exclusive logic circuits 32-1 to 32-4. When they match, the value 0 is supplied to the bit addition circuit 33.

The bit addition circuit 33 calculates the sum of these. This bit addition value indicates the number of unmatched pixels for each scanning line. The sum is output from the binary line matching unit 30 via a signal line and supplied to the multiplexer 34.
The multiplexer 34 sequentially selects the output of the bit addition circuit 33 of each line according to the selection control signal sent from the selection control line, and supplies the output to the adder 35 via the signal line. The sum of the input values is obtained by the adder 35 and the register 36, and the number of mismatched pixels between blocks is obtained as an evaluation function value.

The operation of the comparison detection block 27 will be described. FIG. 4 shows the configuration of the comparison unit of the comparison detection block 27. The comparison unit includes a register 41 for holding the minimum evaluation function value, a register 42 for holding the shift amount of the reference block RB indicating the minimum evaluation function value, and the current minimum evaluation function value and the output of the binary operation unit to be compared this time ( It comprises a comparator 43 for comparing with a block evaluation function value of a certain shift amount, and multiplexers 44 and 45.

In the comparison section, the register 41 holds the minimum value of the evaluation function value at each time, and the register 42 holds the shift amount of the reference block RB corresponding to the minimum value. The comparator 43 compares the evaluation function value supplied via the signal line with the minimum value of the evaluation function supplied from the register 41, and outputs the comparison result to the two multiplexers 44 and 45 via the signal line. Is sent as When the input from the binary operation unit is small, the multiplexer 44 uses the input evaluation function value from the binary operation unit
1, and the multiplexer 45 updates the contents of the register 42 with the input shift amount from the binary operation unit. Finally, the shift amount held in the register 42 is sent to the majority / integration determination unit 28 as the motion vector of the detection block KB.

The operation of the majority / integration determination section 28 will be described. Here, a process of calculating a detection block in which no motion has been detected by comparing the detection block of each R, G, B component with the reference block, that is, an indefinite block from information of surrounding known blocks, and processing of each R, G, B A process of integrating and calculating one motion vector from the motion vector information of the component detection block is performed.

The motion vector information detected by the motion vector detection unit 25 for each detection block KB obtained by dividing the current field into a plurality of parts is input to the majority / integration determination unit 28. The majority / integration determination unit 28 determines the number of motion pixels and the motion direction of each detection block KB by the indefinite block process and the majority process, and registers the determination result in the motion vector table 29.

The indefinite block processing executed by the majority / integration determining section 28 will be described with reference to FIGS.

When the multi-level binary image is divided for each block, the number of motion pixels can be obtained if the edge portion is within the block. When the regions are adjacent to each other (the region indicated by the symbol F in FIG. 13), the number of motion pixels in the region of the block cannot be found.

In such a case, when detecting a motion vector, the number of motion pixels in this block area is undefined (for example, a flag is set to 1) and temporarily registered separately from other blocks. Then, the number and direction of motion pixels of the undefined block FT area sandwiched between the known block KT areas are determined by the known block KT.
Ask from. The undefined block FT area is a known block KT.
Since the motion is the same as that of the region, the same value as the number of motion pixels of the known block KT surrounding these is adopted.

As this method, an image map method can be considered. For example, in the case of a so-called VGA type having a binary image size of 640 horizontal pixels and 480 vertical pixels, if the motion vector detection block KB is 16 * 16 pixels, the number of blocks is 40 horizontal and 30 vertical as shown in FIG. Can be divided into a total of 1200 block areas. For this reason, 40
* From the shape of the known block KT surrounding the undefined block FT as an image map of 30 pixels, the undefined block FT
Can be requested. Here, the indefinite block FT is 2
It is assumed that the value image is an area of data “1” and the known block KT is an area of data “0”.

In this method, a 3 * 3 edge detection window operator shown in FIG. 15 is used. In FIG. 15, 3 *
9 examples are shown among the combinations of 3 edge detection patterns. In this window, a 40 * 30 block image is scanned, and if there is a portion that matches the edge pattern with respect to the point of interest, it is the boundary between the undefined block FT area and the known block KT area.
T is replaced with a known block KT. The number of motion pixels of the undefined block FT is the known block KT including the point of interest.
Of motion pixels.

Next, the linear interpolation method, which is a highly accurate method, will be described. FIG. 16 shows a block relationship including an undefined block according to this method. The procedure of the linear interpolation in this case is performed as follows.

In FIG. 16, first, a known block with a flag '-1' is searched rightward with reference to the block of interest (marked with * in the figure). If this block exists, its motion vector is extracted and the reference block is extracted. Let it be 1. At this time, the motion vector at the position of the reference block 1 is (x1, y
1), and the distance from the target block is d1. Note that the motion vector is determined by the number of motion pixels and the motion direction (+/-)
It is assumed that

Next, a known block with the flag '-1' is searched leftward with reference to the block of interest, and if this block exists, its motion vector is extracted and set as a reference block 2. At this time, the motion vector at the position of the reference block 2 is indicated by (x2, y2), and the distance from the target block is d2. Further, a search is performed for a known block with the flag '-1' in the upward direction based on the block of interest,
If this block exists, its motion vector is extracted and set as a reference block 3. At this time, the motion vector at the position of the reference block 3 is indicated by (x3, y3), and the distance from the target block is d3. Lastly, a known block with the flag '-1' is searched downward with reference to the target block, and if this block exists, its motion vector is extracted and set as a reference block 4. At this time, the motion vector at the position of the reference block 4 is indicated by (x4, y4),
The distance from the block of interest is d4.

As described above, the weight for each reference block is calculated from the distance between the reference block extracted by the search in the horizontal and vertical directions and the target block, and the weight and the motion vector of the reference block are used as follows. A motion vector of the block of interest is obtained according to an arithmetic expression based on linear interpolation.

From FIG. 16, the weight w for each reference block is obtained.
Is given by the following equation.

[0068]

(Equation 1) Then, the motion vector (mx, my) of the target block obtained by linear interpolation is

[0069]

(Equation 2) Becomes

Here,

[0071]

(Equation 3) It is. This is repeated for the number of blocks, and the motion vector of the undefined block is obtained by linear interpolation.

FIG. 17 shows a graphical positional relationship of linear interpolation. Here, i, j, k, l are distances from the point of interest P,
Pi, Pj, Pk, and Pl are i, j, k,
Assuming that the value is a point separated by l, the interpolation formula for the target block position P is as follows.

Assuming that the weight for each point is I,

[0074]

(Equation 4) The value of the point to be obtained (Px, Py) is given by

[0075]

(Equation 5) Becomes here,

[0076]

(Equation 6) It is.

As described above, the motion vector of the undefined block FT area is determined for each of the R, G, B
It is obtained from the number of motion pixels of T by an arithmetic expression. Then, a motion vector value of each block is obtained from the R, G, and B motion vectors by an integration process such as majority decision.

Here, the specific contents of the integration processing by majority decision processing will be described with reference to the flowchart shown in FIG.

Since each block is the same block for the R, G, and B images, the motion direction and the number of motion pixels must be the same. However, a difference may occur due to a binary calculation in a block of the target image and a calculation error of the comparison unit.

The number of motion pixels in the x and y directions for each block is fetched (S1), and it is determined whether the x and y movement directions match for each of R, G and B (S2). In this case, in the movement direction correction / selection (S3), the binary operation of each of the R, G, and B blocks, and the movement directions of x and y output from the comparison detection block are coordinate axes from the first quadrant to the fourth quadrant. Since the reference is assigned to the reference, it is determined whether or not the reference is the same for R, G, and B. This is R, G,
In each of the x direction and the y direction of the movement direction of B, the majority decision is performed including the sign of the movement of the corresponding block and the block adjacent thereto, and registered in the motion vector table 29 as the movement direction (S6). By such means, the accuracy of the movement direction is improved.

Further, the correction / selection of the number of pixels (S5) is similarly performed by the binary operation of each of the R, G, and B blocks, and the number of x and y motion pixels output from the comparison detection blocks 26 and 27 is also R, G
It is determined whether G and B are the same (S4). in this case,
If the numbers of motion pixels of R, G, and B are completely different, an average value of weights that are close to each other and weighted in order from the median of the values is adopted as the number of motion pixels in the motion vector table 29. Register (S6). This average value is usually rounded off.

For example, if the number of motion pixels in the x direction is R, G,
B are 6, 5, and 2, respectively, and their weights are 2, 3,
Let it be 1. The number of motion pixels in the x direction is (2 * 6 + 3 * 5 +
1 * 2) divided by (2 + 3 + 1), that is, 5 pixels.

The same applies to a case where two identical motion pixel values exist for R, G, and B. For example, it is assumed that the number of motion pixels in the x direction is 2, 5, and 2, respectively, and their weights are 3, 1, and 2. Therefore, the number of motion pixels in the x direction is (3 * 2 + 1 * 5 + 2). * 2) divided by (3 + 1 + 2), that is, three pixels. The same applies to the y direction.

Further, an integrated processing method for improving the accuracy of a motion vector will be described below.

FIG. 19 is an operation flowchart for obtaining one motion vector from each of the R, G, and B motion vector values.

In the integration processing shown in FIG.
Due to the arithmetic processing using only the target blocks of the G and B images, there may be a case where the obtained motion vector value does not match the actual value in terms of accuracy. It has been suggested that whether the block of interest is an indefinite block or a known block, and that it is necessary to take into account the motion vectors of the peripheral blocks of the block of interest.

Therefore, as shown in the operation flowchart of FIG. 19, in this integration processing, the flag of the block of interest is first checked (S10), and the flag is set to '1' (S11).
And '0' (S12), the motion vector of the target block is not extracted, and only when the flag is '-1' (S13), the motion vector of the target block is extracted (S1).
4). Thereafter, the motion vectors near the periphery 8 surrounding the target block are extracted (S15), and the extracted motion vectors are rearranged for each of the x and y components.
The median value is set as a representative value (S16).

This is performed up to the number of R, G, and B to be integrated (S17), and then the processing is divided according to the number of representative values. The representative value is indicated for each of the R, G, and B components. At this time, if the number of representative values is one (S18), the motion vector value of the block of interest is flag =-
As 1, the representative value is left as it is (S23). When there are two representative values, such as when the number of target blocks is even, (S1
9) An average value of the two representative values is obtained (S21). As a result of integration, the motion vector value of the block of interest is flag = −
The result of the average value calculation is taken as 1 (S24). Further, when there are three representative values (S20), an average value of the three representative values is obtained (S22). As the integration result, the motion vector value of the target block is set as the flag = -1, and the average value calculation result is obtained (S24). This process is performed up to the total number of blocks (S26), and is set as the motion vector value of the target block.

FIG. 20 shows an example of an integration process for obtaining one motion vector from the values of the motion vectors of the block of interest and its eight neighboring blocks as an example.

FIG. 20A shows an example in which, for each of R, G, and B motion vectors, a target block and its eight neighboring block motion vector values are extracted. At this time, the block marked with a hatched line is the block of interest. The blocks indicated by (-,-) in the figure indicate blocks in an area where no motion vector is detected (flag = 0). FIG.
(B) shows a case where a valid block of a motion vector is extracted from the R, G, and B vectors.

FIG. 20C shows a procedure for rearranging R, G, and B in ascending numerical order. FIG.
(D) shows an example in which the representative values of R, G, and B are calculated. At this time, a procedure is shown in which the median value of the array is selected when the number of data of the motion vector is odd, and the average value of the center 2 is calculated when the number of data is even. At this time, the fraction of the motion vector value is rounded down. Thus, each R,
The representative values of G and B are obtained.

Using this result, an integration process is performed as shown in FIG. That is, when there is one representative value, the value is used as it is, and when there are two representative values, an average value is obtained from the two values. When there are three representative values, an average value is obtained from the three values. In each of the above-described average value calculations, the decimal part is rounded off.

The motion vector result in this case is given by the following equation.

[0094]

(Equation 7) Therefore, the motion vector in the X direction is obtained as “7”, and the motion vector in the Y direction is obtained as “1”.

As a result of the above integration processing, 640 pixels × 4
In the VGA type with 80 pixels, the number of detection blocks is 16 pixels × 16 pixels, so the total number of detection blocks is 1200. The average error obtained by this integration process was improved by about 20% compared to the above-mentioned method, and the variation did not show an extreme motion vector, and a result was obtained in which the overall motion vector was correctly reflected.

As described above, the number of motion pixels obtained for each of the R, G, and B images is integrated into one value for each block.
It is registered in the motion vector table as the number of motion pixels.

Next, the processing contents of the correction processing block 4 will be described.

FIG. 5 shows the configuration of the data correction processing section 4. In the motion vector table 29, motion vector information of each detection block KB in the current field is registered.

FIG. 6 shows the structure of the motion vector table 29. FIG. 6 shows, for example, 640 horizontal pixels and 48 vertical pixels.
9 shows an example of a motion table in the case of a VGA type capable of displaying 0 pixels. In this case, the detection block is horizontal 1
If there are 6 pixels and 16 vertical pixels, the block is divided into a total of 1200 blocks, and the number of motion pixels and the direction of each block are tabulated. Here, the upper left of each block is the origin of the coordinates. The block number of the detection block KB, the offset value from the origin of the detection block KB, the number of motion pixels from the previous field of the detection block KB (including the motion direction) in the order of the detection blocks in which the motion detection has been performed for the current field image. ) And have registered.

FIG. 7 shows a partial configuration of the correction table 53, and FIGS. 8 and 9 show specific table configurations. The correction table 53 registers the correction data by dividing the movement direction of the detection block KB from the first quadrant to the fourth quadrant. In the first quadrant, each correction data is registered according to the number of motion pixels in the x direction (+ direction), which is the horizontal direction, and in the second quadrant, the number of motion pixels in the y direction (+ direction), the vertical direction. Each correction data is registered according to. In the third quadrant, each correction data is registered in accordance with the number of motion pixels in the horizontal x direction (−direction). In the fourth quadrant,
Each correction data is registered according to the number of motion pixels in the y direction (−direction) which is the vertical direction.

FIG. 8 shows that the correction data for the movement in the x direction in the + and-directions, that is, the movement in the first and third quadrants, is a matrix table corresponding to the change in the signal level and its position. ing. Similarly, FIG. 9 shows that the correction data for the movement in the y direction in the + and-directions, that is, the movement in the second and fourth quadrants, is a matrix table corresponding to the change in the signal level and the position. I have. In this description, no correction data is entered.

In the correction processing block 4, the original image data of the current field is inputted to the false contour occurrence level detecting section 51, and the false contour is predicted to be generated by the signal level of each pixel from the current field image. The x and y coordinates of the predicted occurrence point are detected. As described above, since the signal level at which the false contour occurs is near the 2Nth signal level boundary, pixels near the 2Nth signal level boundary are extracted.

All the false contour occurrence predicted points detected by the false contour occurrence level detecting section 51 are detected by the false contour occurrence pixel candidate detecting section 52.
The pixel position which is recognized as a false contour in an actual display according to the visual characteristics is specified from the false contour occurrence predicted points, and is set as a false contour occurrence pixel candidate. In selecting a pixel position that is recognized as a false contour in an actual display, the density of predicted false contour occurrence points is reflected.

In this case, the density of the candidate points can be detected by a 3 * 3 or 5 * 5 window operator centering on the x and y coordinates of the false contour occurrence predicted point. Based on the number of detection points, it is possible to determine whether or not to perform the correction process when one candidate point is isolated.

A false contour actually occurs in a false contour occurrence pixel candidate when the detection block KB to which the false contour occurrence pixel candidate belongs is actually moving.

If the detection block KB to which the false contour occurrence pixel candidate belongs is a moving block, the motion vector table 2
9, the detection block number and the number of motion pixels of x and y are extracted and supplied to the correction table 53. The correction amount corresponding to the number of motion pixels is registered in the correction table 53. The correction amount of the correction table 53 is determined in advance by a visual experiment using a PDP device, and is made into a table in accordance with the structure shown in FIGS.

The correction data specified from the false contour occurrence level output from the false contour occurrence pixel candidate detection unit 52 and the number of x and y motion pixels extracted from the motion vector table 29 is extracted from the correction table 53 to perform a correction operation. Part 5
Give to 4. When correcting the original image data, the correction calculating unit 54 adjusts the pixel having the false contour occurrence level and the moving pixel to the visual light amount because the correction data corresponding to the number of moving pixels and the moving direction is given from the correction table 53. Correct the signal level.

As described above, in the configuration of the present invention, the motion detection processing section and the data correction processing section are divided, and each processing section has a subordinate relationship, so that flexible correction processing is possible.

[0109]

As is apparent from the above description, according to the present invention, in a display device that performs gradation display by the subfield method, it is possible to greatly suppress the occurrence of false contours when a moving image is visually followed. A moving image display method and a moving image display device can be provided.

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram of a gradation image display device according to an embodiment of the present invention;

FIG. 2 is a functional block diagram of a motion detection processing unit in the gradation image display device according to the embodiment.

FIG. 3 is a circuit configuration diagram of a binary operation unit in the motion detection processing unit;

FIG. 4 is a configuration diagram of a comparison unit in the motion detection processing unit.

FIG. 5 is a configuration diagram of a data correction processing unit.

FIG. 6 is a configuration diagram of a motion vector table.

FIG. 7 is a configuration diagram of a correction table.

FIG. 8 is a table configuration diagram of a first quadrant and a second quadrant in a correction table.

FIG. 9 is a table configuration diagram of a third quadrant and a fourth quadrant in a correction table.

FIG. 10 is a diagram showing an overall processing flow of the gradation image display device according to the embodiment.

FIG. 11 is a diagram showing a relationship between a multi-layer image and a threshold in a sample image.

FIG. 12 is a conceptual diagram of block matching.

FIG. 13 is a conceptual diagram of a known block KT process.

FIG. 14 is a diagram illustrating a specific example of a known block and an undefined block in the undefined block processing.

FIG. 15 is a diagram showing a specific example of an edge detection window.

FIG. 16 is a diagram showing a block relationship of an undefined block process;

FIG. 17 is a diagram showing a positional relationship of linear interpolation in undefined block processing;

FIG. 18 is a flowchart of majority decision processing.

FIG. 19 is an operation flowchart of an integration process.

FIG. 20 is a diagram showing an operation of an integration process;

FIG. 21 is a diagram showing a luminance ratio of a subfield.

FIG. 22 is a diagram showing the principle of false contour generation in the subfield method.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Video signal processing part 2 A / D conversion part 3 Motion detection processing part 4 Data correction processing part 5 Output processing part 6 Image display part 7 X scan driver 8 Y scan driver 20 Delay circuit 21-1, 21-2 Smoothing filter Block 22-1, 22-2 Multi-level binarization block 23-1, 23-2 Multi-level memory block 24-1, 24-2 Address mapping unit 25 Motion detection block 26 Binary operation block 27 Comparison detection block 28 Majority decision Integration judgment unit 29 Motion vector table 30-1, 30-2 Binary line matching unit 30-3, 30-4 Binary line matching unit 34 Multiplexer unit 35 Adder 36, 37 Register 51 False contour occurrence level detection unit 52 False contour occurrence pixel candidate detection unit 53 Correction table 54 Correction calculation unit

Claims (13)

[Claims] 1. A moving image display method comprising: capturing a motion of a pixel near a signal level where a false contour occurs, and correcting the current image data according to the motion of the pixel. 2. A section in the vicinity of a signal level where a false contour occurs is binarized by a threshold into a current field image and a previous field image, and the binarized images are compared to detect the number of moving pixels and the moving direction of the moved pixel. A moving pixel having a signal level at which a false contour occurs in the current field image is corrected in accordance with the number of moving pixels and a moving direction thereof. 3. A current field binary image is divided into a plurality of detection blocks, a reference area is set in a previous field binary image for each detection block, and a plurality of reference blocks and detection blocks are set in the reference area. 3. The moving image display method according to claim 2, further comprising: evaluating a degree of coincidence with the reference block; and detecting a moving pixel number and a moving direction of a moving pixel from a positional relationship between the reference block having the highest evaluation value and the detection block. . 4. A detection block in which no motion is detected in a comparison between a detection block and a reference block is provisionally registered as an indefinite block, and an indefinite block surrounded by a known block in which motion has been detected is the same moving pixel as the known block. The moving image display method according to claim 3, wherein the number and the moving direction are set. 5. The moving picture according to claim 4, wherein in the processing of the undefined block, the moving pixel number and the moving direction of the detection block are obtained by linear interpolation based on the moving pixel number and the moving direction of the known block surrounding the unknown block. Image display method. 6. The method according to claim 1, wherein the number of moving pixels and the moving direction of each of the R, G, and B color components are detected for each detection block, and the number of moving pixels and the moving direction of the detection block are determined by majority decision. Item 3. The moving image display method according to Item 3. 7. When retrieving the number of moving pixels and the moving direction of the detection block and its neighboring blocks of each color component, rearrangement is performed in descending order of numerical values, and the median value is obtained as a representative value. 7. The moving image display method according to claim 3 or 6. 8. A multi-level threshold centering around a signal level of 2 N or a signal level obtained by combining these signal levels, the current field image and the previous field image are defined as 2 levels.
6. The moving image display method according to claim 1, wherein the moving image is detected for each multi-layer image data. 9. The method according to claim 1, wherein the current field image and the previous field image are smoothed before binarization.
The moving image display method according to claim 8. 10. The correction data corresponding to the number of moving pixels and the moving direction corresponding to the signal level at which the false contour occurs is registered in a correction table, and based on the detected number of moving pixels and the moving direction of the moving pixel. 9. The moving image display method according to claim 1, wherein correction data is extracted from the correction table. 11. While registering the number of moving pixels and the moving direction detected for each detection block in a motion vector table,
The coordinate data at the signal level at which a false contour occurs is detected from the current field image to extract a correction candidate pixel, and the number of moving pixels and the moving direction of the detection block to which the correction candidate pixel belongs are extracted from the motion vector table and extracted. 11. The moving image display method according to claim 10, wherein correction data corresponding to the number of moving pixels and the moving direction and the false contour occurrence signal level are extracted from the correction table. 12. The moving image display method according to claim 11, wherein a correction candidate pixel is determined based on a detection density of a signal level at which a false contour occurs. 13. A binarization processing means for binarizing a current field image and a previous field image with a threshold value near a signal level at which a false contour occurs, and comparing a binarized image between fields to determine a pixel movement. A motion detection means for detecting, a correction table in which correction data corresponding to the number of moving pixels and a moving direction are registered for each signal level at which a false contour occurs, and pixels at a signal level at which a false contour occurs from a current field image are extracted A moving image display apparatus comprising: an extracting unit that performs a motion detection result of the extracted pixel and correction data corresponding to a signal level from the correction table to correct the pixel.