JPH0448030B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a motion detection circuit for a video signal, which is required for noise removal from a video signal, data compression of a video signal, and the like.

"Background technology and its problems" There are two methods for detecting motion in television images: dividing the screen into a large number of blocks and matching each block between two consecutive frames; A method is known in which motion is detected from the relationship between a frame difference signal between frames, an inter-pixel difference signal, and an inter-line difference signal in the current field. In the former method, it is necessary to increase the block size to some extent in order to improve the matching accuracy, but this inevitably lowers the resolution. The latter method requires complicated arithmetic processing and is impractical, and there is a risk that the background appearing from the shadow of a moving object, called uncovered background, may be mistakenly detected as a moving pixel.

As shown in FIG. 1, if an object 31 in the television image moves in the direction indicated by the arrow, the background pixels included in the shaded area 32 in FIG. 1 will be erroneously detected as moving pixels. Ru. At the same time, among the parts that overlap with the movement of the previous frame, the part with the same signal level, that is, the motion area 33 with a uniform level, is not detected as motion, and only the pixels included in the dotted line area 34 are detected as moving pixels. There was a problem with this.

``Object of the Invention'' Therefore, an object of the present invention is to provide a motion detection circuit that does not erroneously detect an uncovered background as a moving pixel and can also detect a region of motion with a uniform level. There is a particular thing.

Another object of the present invention is to provide a motion detection circuit with better resolution than the block-by-block matching method.

Still another object of the present invention is to provide a motion detection circuit that can follow screen changes such as scene changes and can also remove colored noise.

"Summary of the Invention" This invention focuses on the fact that a television image 37 is constructed by placing a moving image 36 in front of a fixed background image 35, as shown in FIG. This method attempts to extract a background image from an input digital video signal and detect a moving image based on this background image.

The present invention provides a first comparing means for comparing pixel data of a digital video signal and background pixel data stored in a first frame memory corresponding to the pixel data; and a second comparison means for comparing the previous frame pixel data stored in the corresponding second frame memory, and when the comparison output of the first comparison means is less than a predetermined value and the first
The background pixel data is updated by the input pixel data when the comparison output of the first comparison means is greater than a predetermined value and the comparison output of the second comparator is less than a predetermined value, and the comparison output of the first comparison means is is a predetermined value or more and the comparison output of the second comparing means is a predetermined value or more, the video signal motion detection circuit detects input pixel data as motion pixel data.

"Embodiment" An embodiment of the present invention will be described below with reference to the drawings. In FIG. 3, 1 is an input terminal for a video signal, and 2 is an input terminal for a timing reference signal of this input video signal. The video signal is
The timing reference signal is supplied to the A/D converter 3, and the timing reference signal is supplied to the address control circuit 4, and the sampling block formed by the address control circuit 4 is connected to the A/D converter.
The A/D converter 3 generates a digital video signal in which, for example, one sample is converted into 8 bits.

This digital video signal is sent to the frame memory 5, subtraction circuit 8, subtraction circuit 9 and AND gate 1.
0. Frame memory 5 has 1 pixel 8
Bit data for one frame is stored, and a subtraction circuit 8 subtracts pixel data corresponding to the same sampling position of the previous frame from the input pixel data. The difference output of the subtraction circuit 8 is converted into an absolute value by the absolute field conversion circuit 11 and is supplied to the comparison circuit 12. A threshold value is supplied to the comparator circuit 12 from a terminal 13, and a judgment output is generated which becomes 1 when the difference output is less than the threshold value and becomes 0 when the difference output is greater than or equal to the threshold value. It is supplied to the weighting coefficient control circuit 14.

Reference numeral 6 indicates a background memory in which one frame of background pixel data is stored, and reference numeral 7 indicates a weighting coefficient memory in which one frame of weighting coefficients is stored. Background memory 6 stores 12-bit background pixel data, and weighting coefficient memory 7 stores 3-bit weighting coefficients. Frame memory 5, background memory 6
and the weighting coefficient memory 7 are commonly controlled by the output of the address control circuit 4, so that the same address is specified.

Background pixel data at the same sampling position is subtracted from the input pixel data by a subtraction circuit 9, and the difference output of this subtraction circuit 9 is supplied to an absolute value conversion circuit 15. The absolute value of the difference from the absolute value conversion circuit 15 is supplied to the comparison circuit 16 and compared with the threshold value from the terminal 17, and becomes 1 when the difference output between the input pixel data and the background pixel data is less than the threshold value, A determination output of 0 is generated when the difference output is equal to or greater than the threshold value. This determination output is supplied to the weighting coefficient control circuit 14.

18 is a multiplication circuit, and the output of the subtraction circuit 9 is multiplied by a weighting coefficient α. This weighting factor α is generated from the weighting factor control circuit 14. The weighting coefficients from the weighting coefficient control circuit 14 are written into the weighting coefficient memory 7, and the weighting coefficients read from the weighting coefficient memory 7 are supplied to the weighting coefficient control circuit 14. The product output of the multiplier circuit 18 is supplied to the adder circuit 19, where it is added with the background pixel data read out from the background memory 6, and this addition output is written into the background memory 6, and the AND
The signal is supplied to the game 20.

The weighting factor control circuit 14 includes comparators 12 and 16
A detection signal is generated for each pixel (0 for a moving pixel, 1 for a background pixel), and this detection signal is ANDed.
The output of the inverter 21 is supplied to the AND gate 10. Therefore, output terminal 2 of AND gate 10
2, motion pixel data is selectively extracted from the input pixel data, and background pixel data is selectively extracted from the output terminal 23 of the AND gate 20. A moving object is extracted by the outputs appearing at these output terminals 22, 23, and 24. Furthermore, each extracted moving object is compared with past moving objects that have already been obtained, and after that, data compression, noise removal, Y/C separation, image recognition, etc. are performed according to the application purpose. , processing is performed for each of the moving area and fixed area.

In one embodiment of the invention described above, the background memory 6, the subtraction circuit 9, the multiplication circuit 18 and the addition circuit 1
9 constitutes a digital filter using the background memory 6 as a one frame delay element. That is, if the weighting coefficient is α, the input pixel data of the k-th frame is Zk, and the background image data read from the background memory 6 is X _k-1 , the background image data (estimated value) of the _k -th frame is , is expressed by the following equation.

X _k = α・(Z _k −X _k-1 ) + X _k-1 = (1 − α)・X _k-1 + α・Z _k In this example, when the background does not change,
For example, α is fixed at 1/16. White noise contained in the input image is removed by repeating the calculation expressed by the above equation over a plurality of frames, and the S/N of the background pixels stored in the background memory 6 is improved. Also, when the background changes due to scene changes, etc., the weighting coefficient α is increased from 1/16 to twice every frame to shorten the response time and avoid the influence of colored noise. (1/16→1/8→1/4→1/2→
1) and a weighting coefficient α that increases exponentially,
I am trying to update the background image gradually.
Therefore, there are five types of weighting coefficients α, each of which is expressed by three bits. Set the weighting coefficient α to 2
The reason why the value is set to a power is that the multiplication circuit 18 is realized by a shift register or a selector.

In one embodiment of the present invention, after passing through the initial state, a background image is stored in the background memory 6, and a moving pixel detection operation as shown in the flowchart of FIG. 4 is performed.

First, the subtraction circuit 9 and the absolute value conversion circuit 15 detect the absolute value of the difference between the current pixel and the background pixel. The absolute value of this difference and the threshold value are compared by the comparison circuit 16, and when the absolute value of the difference is less than the threshold value, branch 1 processing is performed. Branch 1 is made by determining that the current pixel belongs to the background image, and the weighting coefficient α is fixed at 2 ⁻⁴ (=1/16). Then, the output of the subtraction circuit 9 is multiplied by this weighting coefficient α, added to the readout output of the background image memory 6 by the addition circuit 19, and written to the same address in the background image memory 6, thereby updating the background pixels. . The weighting coefficient α at this time is written into the weighting coefficient memory 7.

When the absolute value of the difference between the current pixel and the background pixel is equal to or greater than the threshold value, the absolute value of the difference between the current pixel and the corresponding pixel of the previous frame detected by the subtraction circuit 8 and the absolute value conversion circuit 11 is equal to the threshold value. It is checked whether the comparison output of is below the threshold value or above the threshold value. When the output of this comparison circuit 12 is less than the threshold value,
Branch 2 processing is performed, and when the threshold value is exceeded, branch 3 processing is performed.

Branch 2 processing is performed by scene change etc.
This is what happens when the background changes. First, the weighting coefficient α (=
1/16) is read out and multiplied by this weighting coefficient α by the multiplier 18. Therefore, the background pixel update made in this frame is (X _k = 15/16
This is similar to branch 1, which is represented by X _k-1 +1/16Z _k ). When the background pixel is updated, α is doubled in the weighting coefficient control circuit 14, and it is determined whether this doubled new weighting coefficient is larger than 1. If it is smaller than 1, This doubled weighting coefficient α is stored in the weighting coefficient memory 7. When starting from a weighting coefficient of 1/16, the weighting coefficient α changes (1/8 → 1/4 → 1/2 → 1) every frame.

When the background is switched, the weighting coefficient α changes as described above, and the processing does not proceed to the background pixel update process of branch 1 until five frames later when (α=1). Therefore, even if colored noise is mistakenly detected as a new background pixel and processed in branch 2, the correct background pixel will be saved. The manner in which the weighting coefficient α changes is set so as to improve responsiveness when updating background pixels and to remove colored noise.

Branch 3 is when the current pixel is determined to be a moving pixel, and in this case, the background memory 6 is not updated. Once moving pixels have been detected as described above, the next step is to extract moving objects.

"Application example" In the case of a digital color video signal, it is divided into component signals of a luminance signal and two color difference signals, and the moving pixels of each color difference signal are detected using the detection results of moving pixels obtained for the luminance signal. Just do it like this.

[Effects of the Invention] Compared to the method of matching each block, the present invention enables detection of moving pixels pixel by pixel and improves resolution. Moreover, unlike the case where the difference output between frames is visible, the present invention can prevent erroneously detecting the uncovered background as a moving pixel, and can also detect motion at a uniform signal level. Furthermore, the present invention can follow changes in the screen such as scene changes.

[Brief explanation of the drawing]

FIG. 1 is a schematic diagram used to explain a conventional video signal motion detection circuit, FIG. 2 is a schematic diagram used to explain the present invention, FIG. 3 is a block diagram of an embodiment of the present invention, and FIG. The figure is a flowchart used to explain the operation of an embodiment of the present invention. 1... Video signal input terminal, 5... Frame memory, 6... Background memory, 7... Weighting coefficient memory, 8, 9... Subtraction circuit, 12, 16... Comparison circuit, 14... Weighting coefficient control circuit.

Claims (1)

[Claims] 1. A first comparison means for comparing pixel data of the digital video signal and background pixel data stored in the first frame memory corresponding to this pixel data, and a first comparison means for comparing pixel data of the digital video signal and background pixel data stored in the first frame memory, and a second comparison means for comparing the previous frame pixel data stored in a second frame memory, and when the comparison output of the first comparison means is less than a predetermined value and the first comparison means When the comparison output of the second comparator is greater than or equal to a predetermined value and the comparison output of the second comparator is less than a predetermined value, the background pixel data is updated by the input pixel data, and the comparison output of the first comparison means is A motion detection circuit for a video signal, wherein the input pixel data is detected as motion pixel data when the comparison output of the second comparing means is greater than or equal to a predetermined value.