JPH11110565A - Motion detector - Google Patents

Abstract

PROBLEM TO BE SOLVED: To detect accurate motion for all kinds of frame images by performing a binarization processing through variable threshold values, based on three pixel data within the frame image for a prescribed pixel under consideration and discriminating whether or not it is a motion pixel corresponding to a binary image based on the binarization processing. SOLUTION: For inputted video signals Pi, image data for one frame among the image data D which are digital data by an A/D converter 101 are inputted to a frame memory 102 by the control of a controller 103. The frame memory 102 stores the pixel data for constituting a first field among the inputted image data for one frame in an odd-numbered row address and the image data for constituting a second field in an even-numbered row address in correspondence. Then, a CPU 104 reads the input image data of three points from the frame memory 102, performs binarization processing based on the image data, obtains the binary image and discriminates whether or not it is a motion pixel.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image motion detecting apparatus for a video tape recorder, a high-definition television, and the like capable of displaying a still image by using an interlaced scanning method.

Further, the present invention relates to an image output system including a video printer system for printing a still image from a continuous image such as a video image using an interlace scanning method.

[0003]

2. Description of the Related Art Video tape recorders, high-definition televisions, and the like generally employ an interless scanning system and have a still image display function. Recently, an image output device such as a video printer can be connected to a video tape recorder, a high-quality television, or the like, and a still image is printed from a continuous image.

Here, the interlaced scanning is such that one frame image is divided into two field images having different times, and the second scanning fills the path of the first scanning, that is, odd scanning and even scanning. This is a scanning method of projecting an image by performing scanning twice, and two images (field images) projected twice are converted into one image (field image) due to an afterimage effect on human visual characteristics. Frame image).

However, in a device to which interlaced scanning is applied, when a still image is created by directly converting the two field images sent in two times into a frame image, for example, the camera and the subject are completely There is no problem if it is stationary, but if any of them is moving, the continuity of the image between the odd scan lines and the even scan lines of the image is lost due to the time lag between the field images, A shift occurs between the image of one field and the image of the second field, and as a result, the quality of the displayed image is reduced. When the image quality of a still image is reduced, the image output result of a video printer or the like is similarly low in quality. Since the display and printing of a still image can be similarly discussed, the following description will be focused on the display.

Therefore, for example, when a still image is displayed on a general video tape recorder, a high-quality television, or the like, 2
A still image is displayed by generating a frame image by interpolating only one of the field images.

However, in this case, since the data of half the scanning lines is interpolated into one frame image, the original information amount is reduced by half, and the image is taken in a state where the camera and the subject are completely still. As compared with the frame image composed of the two field images, the deterioration of the image quality is inevitable.

[0008] Therefore, a still portion in an image is treated as a frame image as it is, and a moving portion in the image is determined by one of the first field and the second field.
It has been practiced to derive a frame image based on a single field image. In this case, it has been necessary to separate a moving part and a stationary part in the image.

Therefore, as shown in FIG. 5, two frame images are used as a method of detecting a moving portion, and the luminance level difference between pixels in the first field and in the second field is obtained, so that each of the moving points is detected. And then taking the logical sum of both.

It is not appropriate to simply take the difference between the first field and the second field because of the positional deviation in the vertical direction due to interlaced scanning. In this case, however, the difference between the first field and the second field is not sufficient. Therefore, it is considered appropriate to compare the difference between the first field and the second field. However, two frame images (four field images) are used. Temporary recording means (memory, etc.)
Is required twice, which causes a problem that hardware cost is significantly increased.

In this case, two adjacent field images among four continuous field images are combined, and a portion of the image detected as a still portion constitutes a frame image. However, if motion detection is performed between every other field image, a further time lag occurs between the field images, and an error may occur in the detection depending on the motion of the image.

In order to solve the above problems, a method described in Japanese Patent Laid-Open No. 6-141283 has been proposed. As shown in FIG. 6, only one frame image is used. Comparing the brightness level of each pixel in the first field (reference video signal) with the brightness level of each pixel in the corresponding second field (motion detection video signal). Is smaller than the threshold value, the pixel is determined to be a stationary portion, and the second field image portion is read as it is. If the luminance level difference is larger than a predetermined threshold, the pixel is determined to be a moving portion and the first field image is interpolated. Output in place of the second field.

In this publication, each pixel of the second field image corresponding to each pixel of the first field image is referred to as a pixel of interest (i row and j column) of the first field image as shown in FIG. There is a description that the average value of two pixels at the (i-1) row and j column and the (i + 1) row and j column of the second field image adjacent to the upper and lower positions can be set.

[0014] Furthermore, it is disclosed that a plurality of pixels adjacent in a field are collectively formed into a block, and the absolute sum of the difference in luminance level of the corresponding pixel between the blocks is calculated and compared with a predetermined threshold value. ing.

[0015]

Problems to be Solved by the Invention
According to the method described in Japanese Patent Application Publication No. JP-A-2003-163, only one frame image is used to perform a motion determination, and a luminance level difference between a certain pixel of the first field image and a corresponding pixel of the second field image is calculated. If the luminance level difference is smaller than a predetermined threshold value, it is determined to be a still portion and the second field image is output as it is, and if larger than the predetermined threshold value, the first field image is interpolated and the second field image is interpolated. In the case where the predetermined range is a still image, the image is displayed as it is, so that the resolution can be prevented from deteriorating. ) Can be displayed while still images can be displayed, and an increase in hardware cost can be minimized.

The difference in luminance level between pixels to be compared is
There is disclosed an idea of preventing a false determination by determining a stationary portion when the value is smaller than a predetermined threshold value.

Further, a luminance level difference is obtained between the corresponding blocks, and when the sum of the absolute values does not exceed a certain threshold value, it is determined that the block is a stationary portion, so that the influence of noise on the luminance level difference in a minute range is reduced. There is a statement that it will be difficult to receive.

However, in the prior art using one frame image, accurate motion detection is not always performed. For example, a pixel (X in the first field)
(0)), the motion detection is performed by comparing the difference between the average value of the upper and lower two adjacent pixels (X (−1), X (+1)) of the second field with a predetermined threshold value. Considering this, for a sharp edge portion as shown in FIG. 8, X (0)
It is expected that the respective values of and (X (-1) + X (+1)) / 2 have a considerable difference in calculation. That is, X (0) is X
Despite the fact that the density is close to either (-1) or X (+1), the difference between the average value of X (-1) and X (+1) is somewhat large. Only by comparing with a fixed predetermined threshold value, a mere edge portion as described above is erroneously determined to be a moving portion, and accurate motion detection cannot be performed.

Therefore, despite the motion detection processing for treating a still portion as a frame image, an edge portion having a high importance as an image is erroneously detected. While the image is treated as a frame image based on the motion determination result, the edge portion is erroneously detected as a motion portion and is interpolated in the field. There is no significant difference above, and there is a problem that the improvement in image quality can hardly be expected.

Also, for example, in an actual blurred portion in an area where the density difference from the background is small, the luminance level difference between the target pixel and the comparison pixel becomes a relatively small value, but erroneous determination due to noise or the like is prevented. If a large value is set for the fixed predetermined threshold value to perform
As a result, there is a problem that a moving part is erroneously determined to be a stationary part.

In the same publication, a plurality of pixels adjacent to each other are collectively divided into blocks, and the sum of absolute values of the luminance level differences of the corresponding pixels between the blocks is compared with a predetermined threshold value. Although it is described that difficult motion detection can be realized, even in this case, for example, if there is a slight movement in the block area, the luminance level difference of each pixel is canceled in the block, and the absolute value sum is set to a predetermined value. It may be less than the threshold value, and as a result, detection failure may occur.

Furthermore, when motion is detected for each block, the processing time for motion detection becomes longer, and a separate storage means for temporarily storing pixel data for the block is required, which complicates the hardware configuration and increases the manufacturing cost. There are also problems such as an increase in cost and a complicated processing program.

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems, and has been made in consideration of a motion which can accurately detect a motion of any kind of frame image with a simple algorithm or a small hardware configuration. It is an object to provide a detection device.

[0024]

According to a first aspect of the present invention, there is provided a motion detecting apparatus for detecting a motion area in an image from a frame image composed of two field images of an odd field image and an even field image by interlaced scanning. In the apparatus, the motion detection device is configured to determine three pixels, a target pixel, an upper adjacent pixel of the target pixel, and a lower adjacent pixel of the target pixel, in the frame image for a predetermined target pixel to be determined as a motion point. It has a binarization processing unit for performing a binarization process based on a variable threshold based on data, and a motion pixel determination unit for determining whether or not a pixel is a motion pixel according to a binary image based on the binarization process. .

According to the motion detecting device of the first aspect, the motion can be detected with extremely high accuracy.

That is, in the present invention, the threshold value can be variably set based on the pixel data such as the luminance level of the three pixels (the target pixel and its upper and lower neighboring pixels). The threshold value can be compared with the corresponding variable threshold value, and an appropriate threshold value can be appropriately set for each image of interest.

Since the threshold value is not a fixed value, it is not necessary to perform motion detection for each image block in order to avoid the influence of noise, and the processing time for motion detection becomes longer, and the processing program and hardware configuration become complicated. It never happens.

Furthermore, motion detection is performed by performing binarization processing in accordance with a binary image based on a variable threshold, so that an edge portion of an image which was erroneously detected as a moving portion in the prior art is detected as a moving portion. Never do. That is, in the related art, when the threshold value is set low, there is a problem that erroneous detection of a stationary edge portion increases, which is extremely advantageous.

In the same manner, the binarization based on the binarization process is performed.
Since it is determined whether or not the pixel is a motion pixel according to the value image, the motion can be detected even in a region where the background density difference is small, that is, a small change in the density difference, and the motion detection omission of a blurred portion is eliminated.

In the prior art, there is a two-frame type in which the difference between fields of the same kind is used for motion detection. However, in the present invention, since the one-frame type is used, the number of used frame memories can be reduced to one. There is no cost increase.

According to a second aspect of the present invention, there is provided a motion detecting apparatus.
Wherein the binarization processing means performs binarization using an average value of three pixel data of a target pixel, an upper adjacent pixel of the target pixel, and a lower adjacent pixel of the target pixel as a threshold value.

That is, according to the second aspect of the invention, 3
With two pixel data, binarization can always be performed without losing information on the moving part of the image. In addition, complicated processing is not required to set the threshold, and extremely good motion detection is achieved by a relatively simple apparatus configuration and processing method. The result can be obtained.

Therefore, it is possible to provide a motion detection device capable of performing accurate motion detection for all types of frame images with a simple algorithm or a small-scale hardware configuration.

[0034]

Embodiments of the present invention will be described below in detail with reference to the drawings. In particular, FIG. 1 is a block diagram showing a configuration of a motion region detection device according to an embodiment of the present invention, FIG. 2 is an explanatory diagram showing the vicinity of a pixel detected by the motion region detection device, and FIG. FIG. 4 is a diagram illustrating a frame image obtained by directly synthesizing the field image of FIG. 4, and FIG. 4 is an explanatory diagram of interpolating and synthesizing the frame image using only one field in a moving part.

First, an image motion detecting apparatus according to an embodiment of the present invention will be described with reference to FIGS.
As shown in FIG. 1, an image motion area detection device 100 according to the present embodiment receives an input video signal Pi, digitizes the input video signal Pi, and generates an image data D by an A / D converter (analog). digita
l converter) 101, a frame memory 102 for storing image data D for one frame obtained by digitization by the A / D converter 101, and a CPU (Central Processing) described later.
A / D converter 10 under the control of Unit 104
1 and the image data D digitized by the controller 103 for controlling the operation of the frame memory 102 and the A / D converter 101 and stored in the frame memory 102.
And a CPU 104 for executing operations such as motion detection determination and controlling the operations of the frame memory 102 and the controller 103.

For example, the CPU 104 has a RAM (random access memory) not shown.
y) and a ROM (read
In the above description, an example in which a memory having a built-in only memory is used is described.
104 may be externally attached.

In FIG. 2, an image motion detecting device 100
Is a single field image, that is, "..., i-1, i +
The pixel data of the other field image, that is, the pixel data of the row of “…, i−2, i, i + 2,...” Is determined based on the pixel data of the row of “1,. Here, a case will be described where it is determined whether the pixel data of the pixel on the i-th row and the j-th column shown in FIG.

First, in FIG. 1, an input video signal Pi is converted into one frame of image data D of digital image data by an A / D converter 101 under the control of a controller 103.
2 is input.

The frame memory 102 associates, for example, pixel data constituting the first field with odd-numbered row addresses, and image data constituting the second field with even-numbered row addresses. Store.

Next, the CPU 104 shown in FIG. 1 reads X (-1), X (0), X
The input image data of three points (+1) is read. Where X
(-1), X (0), and X (+1) represent positions [j, i-1] in the frame image as shown in FIG.
This corresponds to pixel data existing in [j, i] and [j, i + 1].

Here, three pixels X (−1), X
The average value p of (0) and X (+1) is calculated inside the CPU 104. If represented by the equation, the average value p is expressed as follows.

P = | X (−1) + X (0) + X (+1) | / 3 and p, X (−1), X (0), X
(+1) is compared to perform binarization, and the results are Y (-1), Y (0), and Y (+1), respectively. That is,
If X (n)> p, set Y (n) = 1, and if X (n) ≦ p, set Y (n) = 0 (where n is any of −1, 0, and +1). As described above, the partial image composed of three pixels is binarized while keeping the information for motion determination.

In the above method, three methods are used as the binarization method.
We use a method that uses the average value of two pixels as the threshold,
The method is not particularly limited as long as the adaptive binarization is performed with reference to the average density in the vicinity. For example, the position [j,
i-1],..., [j, i + 1] may be increased in the horizontal and vertical directions, and the average value thereof may be referred to.

Next, these Y (-1), Y (0),
A logical operation is performed on Y (+1) to determine whether it is a moving part. It is considered that a moving part is that X (-1) and X (+1) are almost the same pixel data, while X (-1) and X (+1) are almost the same pixel data because a time difference between two fields causes discontinuity between scanning lines. It is considered that only 0) is different.

That is, assuming that one scanning line corresponds to one dot of a pixel, if the image has motion,
If a frame image is formed by synthesizing two field images as they are, the image will be shifted for each scanning line as shown in FIG. 3 due to the time difference between the fields. Therefore, the moving part is adjacent to the first field and the second field. This is considered to be a part where the luminance data and the like of the pixels change alternately.

Therefore, if represented by a binary image, Y (−1) =
1, Y (0) = 0, Y (+1) = 1, and Y
Any of the cases of (-1) = 0, Y (0) = 1, and Y (+1) = 0 is considered to be the end of the moving part, and therefore X is satisfied only when these two conditions are satisfied. (0)
Is the moving point, otherwise it is the stationary point.

In the case of the edge portion of the image, for example, in the case of FIG. 8, when binarization is performed by the above-described method, X (−1) = 0, X
Since (0) = 0 and X (+1) = 1, the above-described motion determination condition is not met, and the motion is not determined, so that the accuracy of detection of the motion is greatly improved.

By adding an appropriate value to the threshold value p, when X (0) is slightly different from X (-1) and X (+1), it can be ignored as a moving part. It is also possible to prevent erroneous detection as a moving part due to noise or the like.

As described above, since the grayscale image is converted into a binary image, and the motion portion is determined by logical judgment, the motion detection can be realized in a simple hardware manner at low cost. Erroneous determination of an edge portion can also be prevented.

Further, if the entire image is also subjected to a fixed threshold value in the binarization process, image information is often lost. However, in the present invention, the threshold value is determined based on the target pixel, its upper adjacent pixel, and its lower adjacent pixel. Since the threshold value can be made variable for each pixel of interest, for example, by selecting the average value of three points around the pixel of interest as the threshold value, the entire image Can be binarized, and as a result, detection omission of a moving portion is almost eliminated.

Using this motion detection, for example, a frame-free still image can be constructed from two moving field images.

As described above, when there is a motion in an image, if two frame images are combined as they are to form a frame image, an image shifted for each scanning line as shown in FIG. 3 due to the time difference between the fields. I will.

In order to avoid this, for a moving part, the frame image may be interpolated and synthesized as shown in FIG. 4 using only one of the two fields.

When this embodiment is applied to the above-described frame image synthesis, if the target pixel X (0) is determined to be a moving point by the method of the present invention, X ( Interpolation synthesis may be performed using X (-1) and X (+1) in a field different from X (0) without using (0).

That is, when X (0) is a moving point,
[J, i] pixel = (X (−1) + X (+1)) / 2, X
When (0) is a stationary point, [j, i] pixel = X (0). By doing so, for example, the image shift portion (blurring) at the end of the moving object in FIG. 4 is replaced with the average value of the upper and lower adjacent images, so that the image has a shift-free edge.

The above calculation process is executed by the CPU 104.

Similarly, when the pixels to be processed are moved, and all the pixels in the..., I−2, i, i + 2,. A still image of high quality is generated.

As described above, the present invention is not limited to the embodiment described above and shown in the drawings, and it is needless to say that the present invention can be appropriately modified and implemented without departing from the scope of the invention.
For example, C that constitutes the motion detection device 100 of the present embodiment example
The PU 104 executes the motion detection determination calculation and the like according to a program stored in the internal ROM. However, the essence of the present invention is not limited to the means for executing the calculation. May be realized by the hardware described above, or the same motion detection processing may be entirely executed by software using a personal computer or a workstation.

In this embodiment, the motion detecting apparatus for displaying a good still image is described. However, for example, a still image is printed from a continuous image such as a video image using an interlace scanning method. It goes without saying that the present invention can be similarly applied to an image output system including a video printer system.

[0060]

According to the first aspect of the present invention, motion detection can be performed with extremely high accuracy.

That is, in the present invention, the threshold value can be variably set based on the pixel data such as the luminance level of the three pixels (the target pixel and its upper and lower neighboring pixels). The threshold value can be compared with the corresponding variable threshold value, and an appropriate threshold value can be appropriately set for each image of interest.

Since the threshold value is not a fixed value, it is not necessary to perform motion detection for each image block in order to avoid the influence of noise, and the processing time for motion detection becomes longer, and the processing program and hardware configuration become complicated. It never happens.

Further, motion detection is performed by performing a binarization process in accordance with a binary image based on a variable threshold, so that an edge portion of an image which was erroneously detected as a moving portion in the prior art is detected as a moving portion. Never do. That is, in the related art, if the threshold value is set low, there is a problem that erroneous detection of a stationary edge portion increases, which is extremely advantageous.

In the same manner, the binarization based on the binarization process
Since it is determined whether or not the pixel is a motion pixel according to the value image, the motion can be detected even in a region where the background density difference is small, that is, a small change in the density difference, and the motion detection omission of a blurred portion is eliminated.

In the prior art, there is a two-frame system in which the difference between fields of the same kind is used for motion detection. However, in the present invention, since the one-frame system is used, the number of frame memories used can be reduced to one. There is no cost increase.

On the other hand, according to the second aspect of the present invention, binarization can be performed without losing information on the moving part of the image, and as a result, a good motion detection result can be obtained.

That is, the three pixel data can always be binarized without losing the information of the moving part of the image. In addition, complicated processing is not required for setting the threshold value, and a relatively simple apparatus configuration and processing method can be used. , An extremely good motion detection result can be obtained.

Therefore, it is possible to provide a motion detection device capable of accurately detecting motion of any type of frame image with a simple algorithm or a small-scale hardware configuration.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of a still image creation device using a motion detection device according to an embodiment of the present invention.

FIG. 2 is a diagram for explaining motion detection by a binarization process centering on a pixel whose motion is detected by the motion detection device according to the embodiment of the present invention;

FIG. 3 is an explanatory diagram illustrating a frame image obtained by directly combining two moving field images.

FIG. 4 is an explanatory diagram of synthesizing a frame image by interpolation using only one field for a moving part;

FIG. 5 is a diagram illustrating a motion detection method using two frame images for explaining a conventional technique.

FIG. 6 is a diagram illustrating a motion detection method using one frame image for explaining a conventional technique.

FIG. 7 is a diagram illustrating an interpolation method of a motion detection method using one frame image for explaining a conventional technique.

FIG. 8 is a diagram illustrating a problem of a motion detection method using one frame image for explaining a conventional technique.

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 motion region detecting device 101 A / D converter 102 frame memory 103 controller 104 CPU

Claims (2)

[Claims] 1. A motion detecting device for detecting a motion region in an image from a frame image composed of two field images of an odd field image and an even field image by interlaced scanning, wherein the motion detecting device is a motion point. Binarization with a variable threshold value based on three pixel data of the target pixel, the upper adjacent pixel of the target pixel, and the lower adjacent pixel of the target pixel in the frame image for a predetermined target pixel to be determined An image motion detection device comprising: a binarization processing unit for performing a process; and a motion pixel determination unit for determining whether or not the pixel is a motion pixel according to the binary image based on the binarization process. 2. The binarization processing means includes a target pixel, an upper adjacent pixel of the target pixel, and a lower adjacent pixel of the target pixel.
2. The apparatus according to claim 1, wherein the binarization is performed using an average value of the pixel data as a threshold value.