JPH0353390A - Outline discriminating device for moving body - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Industrial Application Field] This invention tracks the moving part of the face of a person whose upper body is moving in an image taken with a video camera,
The present invention relates to an apparatus for identifying the outline of a moving body, which is applied when extracting and displaying only a face part.

(Prior Art) FIG. 6 is a block diagram of a circuit example of a conventional contour tracking device for a moving body. In this figure, 1 is a video camera, the output signal of which is digitized by an A/D converter 2, and a luminance signal is extracted by a luminance signal/signal extracting means 3. 4 is a subtractor that calculates a difference signal for each corresponding pixel with respect to the luminance signal of the video signal of two consecutive frames, and the difference signal obtained by the subtractor 4 is sent to a converter 7.
It is binarized by A threshold value for binarization is given by a thousand threshold generation stages 8. Most of the pixels determined by the converter 7 to be larger than the threshold value exist on the outline of the moving part of the video shot by the video camera 1. That is, the pixel group detected by the comparator 7 represents an approximate contour line.

FIG. 7 is another block diagram of the circuit of the conventional contour tracking device for a moving body. This is the blocking means 9 in the circuit of FIG.
．． II1 area block detection means 10 and isolated block removal means 23 are connected. The blocking means 9 divides the binarized pixel group in one frame into adjacent vertical and horizontal n
This is a circuit that divides into blocks each consisting of xn pixels. The moving area block detection means 10 determines how many pixels in each block created by the blocking means 9 have a difference between the luminance signals output from the subtractor 4 than the threshold value given by the threshold value generation means 8. It is determined whether the number is larger or smaller than a predetermined number. If the number is larger than a predetermined number, the block is recognized as a motion area block. The output of the moving area block detection means 1o, that is, the set of moving area blocks exists on a portion of the video shot by the video camera 1 that includes the outline of the moving portion. That is, these blocks represent approximate contours. An advantage of a blocked contour is that it tends to be a continuous area, compared to a contour that is not blocked and tends to be discontinuous.

The output voltage of the subtractor 4 is small for pixels in areas where there is no movement, and a block containing only pixels in areas where there is no movement should not originally be recognized as a moving area block. Recommended area blocks may be detected even in areas where there should be no movement due to vibrations, electrical noise, etc. In this case, the detected moving area block is often isolated. Therefore, when the isolated block removing means 23 is connected after the moving area block detecting means 10, most of the points that appear as noise in areas with no movement are removed, which is effective in improving the S/N ratio.

[KID to be solved by the invention] In the case of the example shown in Fig. 6, most of the points output from the comparator 7 are on the contour, but as mentioned above, points are also output as noise in areas other than the contour. be done. If the threshold value of the converter 7 is increased, the number of noise points will be reduced, but the number of points making up the contour will also be reduced, resulting in the disadvantage that a continuous line will no longer be made up. In other words, it is difficult to obtain clear contour lines. On the other hand, if the threshold value is made smaller, the outline will appear clearer, but there will be more noise points. The range of the optimal threshold is very narrow, and the optimal threshold changes because the difference value changes depending on the intensity of movement, etc., so if you use a constant value as the threshold, it is difficult to obtain a high-quality contour line. is difficult.

In the case of the example shown in FIG. 7, processing to create blocks after binarization is added. A contour part with a high density of points whose difference value is determined to be larger than the threshold by the converter 7 has a high probability of being identified as a moving area block, and a block identified as a moving area has a high probability of being identified as a moving area block. Since points that are not considered to be larger than the threshold value by the comparator 7 are often included, contours tend to appear clearly as continuous regions, while in areas with little movement, the difference value is determined by the comparator 7 to be higher than the threshold value. Since points determined to be larger than the value are likely to be determined to be non-moving areas, there is an advantage that noise is easily reduced. That is, a contour region with good S/H can be obtained. However, forming blocks is equivalent to filling in the outline with the size of the block, and has the disadvantage that only a rugged and thick outline can be obtained.

In FIG. 7, when the 1,000 stages 23 of isolated block removal are connected, the isolated blocks are removed and the S/
N's good contour can be obtained, but the point is that only a rugged and thick contour can be obtained.

The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide an apparatus for identifying the outline of a moving body that can clearly obtain the outline of a moving body.

[Means for solving the problem] Claim (1) of the contour identification device for a moving body according to the present invention
．． The invention described in (2) detects a moving area block, reduces the threshold value only in the area of the block, or in the block and blocks adjacent to the block,
The threshold value for the other parts is increased and the difference signal is again binarized.

[Effect]

Claims related to this invention (1). The invention described in (2) reduces the threshold value only in the blocked moving region where the contour exists, or in that block and the blocks adjacent to that block, and then calculates the difference again.
Since the contour is converted into a value, it is clearly detected as a nearly continuous line. On the other hand, since the difference signal of the area that is not considered to be a moving area due to blocking is binarized again with a larger threshold value, the probability that noise will be detected is reduced.

〔Example〕

FIG. 1 is a block diagram of a circuit showing an embodiment of the invention. Reference numeral 1 denotes a video camera, the output signal of which is digitized by an A/D converter 2, and a luminance signal is extracted by a luminance signal extraction means 3. Reference numeral 4 denotes a subtractor that obtains a difference signal for each pixel corresponding to the luminance signal of the video signal of two consecutive frames, and the difference value obtained by the subtractor 4 is stored in the frame memory 5. Writing and reading of the frame memory 5 is controlled by a write/read control stage 6. On the other hand, the difference signal obtained by the subtractor 4 is binarized by the comparator 7. A set of pixels determined to be significant through binarization constitutes a provisional contour line. However, this provisional contour line is often an unbroken line because the binarization threshold is not completely appropriate. A threshold value for this binarization is provided by threshold generation means 8. This threshold value is a value that is temporarily determined to be an optimal value. 9 is a blocking means, which is a circuit that divides a group of binarized pixels in one frame into blocks each consisting of n×n adjacent pixels in the vertical and horizontal directions; Reference numeral 10 denotes a moving area block detection means, which detects a pixel for which the difference value of the luminance signal output from the subtractor 4 is larger than the threshold value given by the threshold value generation means 8.
The number of blocks in each block created by the blocking means 9 is counted, and it is determined whether the blocks are larger or smaller than a predetermined number.

If the number is larger than a predetermined number, the block is recognized as a work area block. Reference numeral 11 denotes a threshold value setting means, which sets a threshold value smaller than the threshold value generated by the threshold value generation means 8 for pixels in the active area block (contour area), and A threshold value larger than the threshold value generated by the threshold value generation means 8 is set for pixels in the (non-contour area). When setting a small threshold value for contour areas and a large threshold value for non-contour areas, there is a wide range of values that can be selected as the optimal threshold value, so the difference value Contours can be detected stably even if there are variations in the contours. The set threshold is frame memory 1
2.

When the threshold setting is completed, the threshold setting means 11 transmits a signal indicating that the threshold setting is completed to the readout start signal generation stage 13, and the readout start signal generation means 13 receives this signal. write/read control means 6
, 14 read functions are operated in synchronization. A converter 15 binarizes the difference value of each pixel read from the frame memory 5 based on a threshold value read out synchronously from the frame memory 12. The output of the comparator 15 is a line in which part or all of the provisional outline that is the output of the comparator 7 is connected, and is closer to the outline of the moving subject captured by the video camera 1.

FIG. 2 shows groups 16 to 19 of consecutive pixels (small squares filled in black in FIG. 2) which were binarized and determined to be significant by the comparator 7 in FIG. 1, and the blocking means 9.
2 is a diagram showing a moving area block (2×2 four pixel group surrounded by a thick line in FIG. 2) 20.21 detected by the moving area block detection means 10 after being divided into blocks by the moving area block detecting means 10. FIG. Pixel groups 16 to 19 are in blocks 20-1 and 21-
1 and 20-2 and 21-2 are discontinuous. By reducing the threshold values in blocks 20 and 21 and binarizing them again in the above embodiment, new pixels that can be considered significant are generated in these blocks, and pixel group 1
There is a strong possibility that pixel groups 6 and 17 and pixel groups 18 and 19 are continuous pixel groups.

In the above, the threshold value is reduced only for the pixels in the moving area block and binarization is performed again. However, if the threshold value is also reduced for the pixels in the blocks adjacent to the moving area block, the converter 7 obtains The temporary contour lines thus created are easier to connect.

FIG. 3 is a diagram of pixels and moving area blocks in the case where the threshold value is reduced also in blocks adjacent to the moving area block. In FIG. 3, the hatched portion 22 is the pixel area in the block adjacent to the motion area block 20.21. This hatched area 22 includes pixels located between the moving area block 20-2 and the moving area block 21-2. The area between the moving area block 20-2 and the moving area block 21-2 is an area where an outline connecting the pixel groups 17 and 18 exists, and as shown in FIG. If the threshold value is to be reduced including the inner part, pixel groups 17 and 18 in the area between the working area block 20-2 and the moving area block 21-2
There is a strong possibility that a group of pixels forming a contour line connecting the two will be detected. Therefore, as shown in FIG. 3, when reducing the threshold value also in the blocks adjacent to the moving area block, as shown in FIG. Ring detected! IB lines have the advantage of being easy to connect.

FIG. 4 is a block diagram of a circuit according to another embodiment of the invention. The circuit shown in FIG. 4 is the same as the circuit shown in FIG. 1, with isolated block removing means 23 inserted between the dynamic area block detecting means 10 and the threshold setting means 11. The output voltage of the subtractor 4 is small for pixels in areas where there is no movement, and a block containing only pixels without movement is unlikely to be recognized as a primary area block, but lighting, vibration, etc. Due to the influence of electrical noise, etc., moving area blocks may be detected even in areas where there is no movement.

In this case, the detected moving area block is often isolated. Therefore, by removing isolated blocks, a contour line with a better S/N ratio can be obtained than when using the circuit of FIG.

FIG. 5 is a circuit diagram of a further embodiment of the invention. Fifth
The figure shows frame memories 24 and 25 connected to the circuit diagram of FIG. 1. Frame memories 24 and 25 are used to output data from the differentiator 4 while the comparator 15 is performing binarization processing using the difference value stored in the frame memory 5 and the threshold value stored in the frame memory 12. This is a memory for storing the next difference value data and the next threshold data output by the threshold setting circuit 11, respectively. The frame memory 5.12 uses the difference value stored in the frame memory 24 and the threshold value stored in the frame memory 25, and while the comparator 15 is performing the binarization process, the difference unit 4 outputs The next difference value data and the next threshold data output by the threshold setting means 11 are respectively stored.

In the above, a block has been described as a group of four 2×2 pixels, but it goes without saying that blocks may be made up of other numbers of pixels, such as 3×3, 2×3, etc.

[Effect of the invention] As is clear from the above explanation, according to the invention described in claims (1) and (2) of the present invention, the difference values of corresponding pixels of two consecutive frames can be binarized. The threshold value is small only in the area where the outline of the moving body image exists and its vicinity, and is large in other areas, so the outline can be obtained as a continuous clear line, and the threshold value can be obtained in areas other than the outline. This has the advantage of reducing noise. Moreover,
Different thresholds are used for areas where contours exist and other areas, so the range of thresholds is wide enough to stably detect contours even if the magnitude of the difference value changes due to changes in the image condition. There are advantages. Therefore, it becomes easy to select an optimal threshold value regardless of image quality.

Therefore, applying this invention to, for example, video conferencing,
If only the face of a person photographed with a video camera is extracted and displayed on a display device, the outline of the moving part in the video can be clearly detected, making it easier to track the position of the face. To realize a video conference in which the position of the displayed face has little variation and the other party's face is easy to see.

[Brief explanation of drawings]

FIG. 1 is a circuit diagram showing one embodiment of the present invention, FIGS. 2 and 3 are diagrams showing areas where a threshold value is newly set in this invention, and FIG. FIG. 5 is a circuit diagram showing another embodiment of the present invention, and FIGS. 6 and 7 are circuit diagrams showing an example of a conventional contour identification device for a moving body. In the figure, 1 is a video camera, 2 is an A/D converter, 3 is a luminance signal extraction means, 4 is a subtractor, 5 is a frame memory,
6 is a write/read control means, 7 is a comparator, 8
1 is a threshold value generation means, 9 is a blocking means, 10 is a designated area block detection means, 11 is a threshold value setting means, 12 is a frame memory, 13 is a reading start signal generation means, 1
4 is a write/read control means, 15 is a frame memory, 16 to 19 are pixel groups determined to be significant by the converter 7, 20.21 is a moving area block, 22 is a pixel area of a block adjacent to the moving area block, and 23 is a pixel area of a block adjacent to the moving area block. Isolated block removing means 24.25 is a frame memory. Fig. 2 Fig. 3 Image viewing of the block

Claims (2)

[Claims] (1) means for creating a difference signal for each pixel corresponding to two consecutive frames of video signals, a frame memory for storing this difference signal, and means for binarizing the difference signal using a predetermined threshold; Blocking means that divides the pixels having the binarized video signal into blocks each having a predetermined number of pixels; moving area block detection means for counting and determining whether or not the number of pixels is greater than or equal to a predetermined number; a threshold setting means for assigning a threshold value of a value less than or equal to a threshold value, and a threshold value of a value greater than or equal to the predetermined threshold value to an area within the block whose number is less than or equal to a predetermined number; An apparatus for identifying a contour of a moving body, comprising means for binarizing a difference signal of each pixel stored in the frame memory based on a threshold value given by a threshold value setting means. (2) means for creating a difference signal for each pixel corresponding to two consecutive frames of video signals, a frame memory for storing this difference signal, and means for binarizing the difference signal using a predetermined threshold; Blocking means that divides the pixels having the binarized video signal into blocks each having a predetermined number of pixels; moving area block detection means for counting and determining whether or not the number of pixels is greater than or equal to a predetermined number; For areas within the predetermined number of blocks that exist, a threshold value that is less than or equal to the predetermined threshold value is set, and for other areas, a value that is greater than or equal to the predetermined threshold value is set. Threshold setting means for assigning a threshold value; and means for binarizing the difference signal of each pixel stored in the frame memory based on the threshold value assigned by the threshold setting means. What is claimed is: 1. A contour identification device for a moving body, comprising: