JPS6167379A - Detection circuit for movement of television signal - Google Patents

Abstract

PURPOSE:To improve the accuracy of movement detection by providing two independent movement detection systems with respect to horizontal and vertical directions respectively so as to select the movement output of each system. CONSTITUTION:A left side sampling difference DELTAEL is produced by using a subtraction circuit 11 to subtract a picture element S0 before one sample from a present picture element S1 and a right side sampling difference DELTAER is produced by using a subtraction circuit 12 to subtract the present picture element S1 from a picture element S2 after one sample. A subtraction circuit 13 subtracts a picture element S3 before one line from the present picture element S1 to produce an upper side line difference DELTALU and a subtraction circuit 14 subtracts the present picture element S1 from a picture element S4 after one line to produce a lower side line difference DELTALD. The circuit systems for movement detection are provided respectively independently by using in total four slopes, the forward/reverse slope of the present picture element DELTAEL and DELTAER with respect to the horizontal direction and the forward/reverse slopes of the present picture element DELTALU and DELTALU with respect to the vertical direction.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a motion detection circuit for digital television signals.

[Conventional technology]

When digitizing television signals, high-efficiency encoding methods have been considered that reduce the average value of the number of bits per pixel. One of the known methods is an interframe encoding method, and one of the interframe encoding methods is one based on motion compensation. This uses motion detection to find positional relationship information (referred to as motion compensation amount or motion vector) between the current frame and the previous frame, and then manipulates the image of the previous frame based on this motion compensation amount to create a new frame. It is intended to take measures in between.

An example of a motion detection circuit applicable to such an interframe encoding method is described in US Pat. No. 4,278,996. This motion detection is called the gradient method, and as described below, it uses frame differences and slope information (sampling differences in the horizontal direction and line differences in the vertical direction) for all pixels in the motion area. This is to find the amount of movement.

As shown in FIG. 5A, it is assumed that a tilted image with a luminance of about 1 moves in the direction of the arrow (to the right) in one frame in the horizontal direction and has come to the position of the current frame shown in FIG. 5B. .

If the frame difference ΔF is defined as "the pixel of the previous frame subtracted from the pixel of the current frame", then this frame difference ΔF
F is as shown in Figure 5C, and the integrated value of the frame difference ΔF is the area A of the shaded area in Figure 5B.
becomes. This area A is the product of the height h and the amount of movement vl.

Here, if we define the sampling difference ΔE as "the value obtained by subtracting the previous sampling pixel from the current sampling pixel", then
Since the height h is equal to the integrated value of the sampling difference ΔE (FIG. 5D), the magnitude of the horizontal movement amount ■1 can be determined by the following equation.

IVI l=Σ 1ΔFl/Σ1ΔF1 As is clear from FIG. 5C and FIG. They have different signs. Also, unlike the case in Figure 5, in the case of movement to the left,
Integrated value of frame difference ΣΔF and integrated value of sampling difference Σ
ΔE have the same polarity, that is, the same sign.

The direction of movement can be determined from this sign relationship. Therefore, when the sampling difference ΔE is positive (positive slope),
Integration is performed by adding the frame difference ΔF, and when the sampling difference ΔE is negative (negative slope), the area A is obtained by performing integration by subtracting the frame difference ΔF.

In this way, the horizontal motion amount v1 having a direction is determined by the following equation.

vl=Σ(ΔF−sign(ΔF))/Σ1ΔE1 However, the sign (ΔF) becomes 0 when (ΔE=0), and (ΔE≠0
), it becomes 8E/lΔE1.

The above idea can still be used even if extended to two-dimensional motion. In other words, a new frame difference ΔF caused by vertical movement is also added, but since the value of the sign (ΔF) has no correlation with the vertical movement, the horizontal frame difference ΔF caused by vertical movement is The effects of are canceled out. The vertical motion amount v2 can be found by the following equation if the line difference ΔL is defined as "the pixel of the previous line subtracted from the pixel of the current line".

■2-Σ (ΔF・Sign (ΔF))/Σ1ΔLl 1st Δ
FIG. 1 is a block diagram showing the configuration of a conventional two-dimensional motion detection circuit. In FIG. 6, a digital television signal is supplied to an input terminal indicated at 61. This digital television signal is supplied to a frame delay circuit 62 having a delay amount of one frame, a sample delay circuit 64 having a delay amount of one sampling period, and a line delay circuit 66 having a delay amount of one line.

A subtracter 63 subtracts the output of the frame delay circuit 62 (pixels of the previous frame) from the input digital television signal, and calculates the frame difference ΔF from the output of the subtracter 63.
occurs. A subtracter 65 subtracts the output of the sample delay circuit 64 (pixel of the previous sample) from the input digital television signal, and a sampling difference ΔE is generated from the output of the subtracter 65. A subtracter 67 subtracts the pixels of the previous line before the output of the line delay circuit 66 from the input digital television signal, and a line difference ΔL is generated from the output of the subtracter 67.

The frame difference ΔF is supplied to two integration circuits.

One of the integration circuits includes an adder/subtracter 71 and a register 72, and the frame difference ΔF and the output of the register 72 are supplied to the adder/'IJJi adder 71. The other integration circuit is
It consists of an adder/subtracter 81 and a register 82, and the frame difference ΔF and the output of the register 82 are supplied to the adder/subtracter 81. The frame difference ΔF is supplied to adder/subtractors 71 and 81 of these integration circuits.

The adder/subtractor 71 is configured to perform either addition or subtraction operation based on the output of the control circuit 73. The control circuit 73 is supplied with the sampling difference ΔE from the subtraction circuit 65, and when the sign of this sampling difference ΔE is positive, an addition operation is performed, and when the sign of this sampling difference ΔE is negative, a subtraction operation is performed. , a control circuit 73 controls the adder/N adder 71.

The adder/subtractor 81 is configured to perform either addition or subtraction operation based on the output of the control circuit 83. Control circuit 8
3 is supplied with the line difference ΔL from the subtraction circuit 67, and when the sign of this line difference ΔL is positive, an addition operation is performed;
When the sign of this line difference ΔL is negative, the control circuit 83 controls the adder/subtractor 81 to perform a subtraction operation.

The sampling difference ΔE is supplied to a conversion circuit 74, converted to an absolute value, and supplied to one input of an adder 75 for integration. The output of the adder 75 is supplied to a register 76, the output of the register 76 is supplied to the other input of the adder 75, and the integrated value of the absolute value of the sampling difference ΔE is taken out as the output of the register 76.

The line difference ΔL is supplied to a conversion circuit 84, converted to an absolute value, and supplied to one input of an adder 85 for integration. The output of the adder 85 is supplied to a register 86, the output of the register 86 is supplied to the other input of the adder 85, and the integrated value of the absolute value of the line difference ΔL is taken out as the output of the register 86.

As mentioned above, the horizontal movement vl is determined by the frame difference ΔF
It is obtained by adding or subtracting the value of ΔE in the motion region and dividing it by the integrated value of the absolute value of the sampling difference ΔE using the divider 77, and this motion output v1 can be obtained at the output terminal 78. Further, the vertical motion v2 is obtained by adding or subtracting the frame difference ΔF in the motion area by the divider 87 and dividing it by the integrated value of the absolute value of the line difference ΔL, and outputs this motion to the output terminal 88. v2 can be obtained.

[Problem that the invention seeks to solve]

Motion detection using the gradient method described above is based on stationarity (ΔE
Since the assumption is that the continuity of

As shown in FIG. 7A, in the previous frame, pixels a,
b, When there is an extreme value consisting of Since these two signs are the same, it is correctly detected as a movement to the left.On the other hand, when the extreme value shown in Fig. 7A moves to the right by one pixel (Fig. 7c), all pixels In this case, the frame difference ΔF is positive, and the sampling difference ΔE is positive.Therefore, even though the movement is to the right, it is erroneously detected as a movement to the left.

In addition, when the downward extreme value occurs, the frame difference ΔF becomes negative and the sampling difference ΔE becomes negative regardless of whether the direction of movement is left or right. Therefore, even though the true direction of movement is toward the right, it is erroneously determined that the movement is in the direction of two people.

In general television signals, extreme values are everywhere, so a decrease in motion detection accuracy due to the influence of extreme values is a problem that cannot be ignored.

SUMMARY OF THE INVENTION Accordingly, an object of the present invention is to provide a television signal motion detection circuit that can solve the problem of deterioration in motion detection accuracy due to unsteady parts such as extreme values.

[Means for solving problems]

The present invention provides delay and calculation means 7 and 10 for generating a frame difference ΔF between the current frame and the previous frame of a digital television signal, a delay and calculation means for generating a slope ΔE of the digital television signal, and a frame difference an adder/subtracter constituting an integrating means for integrating; a means for dividing the output of the integrating means by the integrated value of the absolute value of the slope ΔE of the digital television signal; State the slope of the digital television signal Δ
At least two motion detection circuits each comprising means for controlling in relation to the polarity of E are provided, and one of the motion detection circuits is configured to detect the current pixel and The other side of the motion detection circuit detects the (approximately ΔEL(ΔLU)) with respect to the previous pixel, and the delay and calculation means 5.12
(2,4° 5.14), the configuration is configured to detect the slope ΔER (ΔLD) between the current pixel and the next pixel, and each of the two motion detection circuits is means 28 (48
), 29 (49).

[Function] Two independent motion detection circuits are provided that use the left side slope ΔEL and the right side slope ΔER as slopes, and the movement detection circuit on the side that correctly detects the direction of movement when the slope is at the extreme value. Selectively retrieve output. Not only in the horizontal direction but also in the vertical direction, the upper slope ΔLU and the lower slope ΔL
Two systems of motion detection circuits using each of D and D are provided independently, and from among the outputs of each motion detection circuit, the one that correctly detects the direction of motion is selected. This selection can be made by comparing the absolute values of the outputs of the two motion detection circuits or based on looking at the signs of the outputs of the two motion detection circuits.

〔Example〕

An embodiment of the present invention will be described below with reference to the drawings.

In FIG. 1, 1 is an input terminal for a digital television signal. This input signal is supplied to a cascade connection of line delay circuits 2 and 3 each having a delay amount of one line, and is also supplied to a sample delay circuit 4 having a delay amount of one sampling period. A sample delay circuit 5 is connected to the output of the line delay circuit 2, and a sample delay circuit 6 is connected to the output of the line delay circuit 3. A frame delay circuit 7 and a sample delay circuit 8 are connected to the output of the sample delay circuit 5.
is connected.

The output of the sample delay circuit 5 is the current pixel to be subjected to motion detection. The output of the sample delay circuit 5 and the output of the frame delay circuit 7 are supplied to a subtraction circuit 10, the corresponding pixel of the previous frame is subtracted from the pixel of the current frame, and the frame difference ΔF is taken out as the output of the subtraction circuit IO.

Assuming that the current pixel is 81, pixels in the vicinity of this current pixel S1 are used for detecting the inclination in the horizontal and vertical directions. However, this tilt detection is performed both between the current pixel and the previous pixel, and between the current pixel and the subsequent pixel. FIG. 2 shows pixels in the vicinity of the current pixel S1.

When the pixel S6 is supplied to the input terminal 1, the output of the line delay circuit 2 is the pixel S2 on the previous line, the output of the line delay circuit 3 is the pixel S5 on the previous line, and the output of the sample delay circuit 4 is the pixel S2 on the previous line. The output is the pixel S4 one sample before.

Therefore, the current pixel S1, which is one sample before the pixel S2, is generated at the output of the sample delay circuit 5, and the pixel SO, which is one sample before the current pixel S1, is generated at the output of the sample delay circuit 8.

Further, a pixel S4 one line after the current pixel S1 is generated at the output of the sample delay circuit 4, and a pixel S3 one line before the current pixel S1 is generated at the output of the sample delay circuit 6.

The horizontal slope is determined by subtraction circuits 11 and 12. That is, in the subtraction circuit 11, the pixel SO one sample before is subtracted from the current pixel S1, and the left side sampling difference ΔEL is generated in the output, and in the subtraction circuit 12,
The current pixel S1 is subtracted from the pixel S2 one sample later, and the sampling difference ΔER on the right side is generated in the output.

The distance in the vertical direction is determined by subtraction circuits 13 and 14. That is, in the subtraction circuit 13, the pixel S3 one line before is subtracted from the current pixel S1, and the upper line difference ΔLtJ is generated in its output, and in the subtraction circuit 14, the current pixel S1 is subtracted from the pixel S4 after the ■ line. and a lower line difference ΔLD is generated at its output.

In these horizontal directions, the values before and after the current pixel (ΔEL
and ΔR, and the slope Δ before and after the current pixel in the vertical direction.
There are four independent circuit systems that perform motion detection using a total of four circuits including LU and ΔLD.

The frame difference Δ■:' obtained at the output of the subtraction circuit 10 is supplied to four first to fourth integration circuits. The first integration circuit includes an adder/remainer 21 and a register 22 to which the output of the adder/remainer 21 is supplied, and the frame difference ΔF and the output of the register 22 are supplied to the adder/subtractor 21. be done. The second integration circuit includes an adder/subtracter 31 and an adder/subtracter 31
The frame difference ΔF and the output of the register 32 are supplied to an adder/subtractor 31. The third integration circuit includes an adder/subtracter 41 and the adder/subtracter 41.
The frame difference ΔF and the output of the register 42 are supplied to the adder/subtractor 41. The fourth integration circuit includes an adder/subtractor 51
and a register 52 to which the output of the adder/subtractor 51 is supplied.
The frame difference ΔF and the output of the register 52 are supplied to the adder/corresponder 51.

The adder/subtracter 21 is controlled by the output of the control circuit 23 to perform either addition or subtraction operation. The control circuit 23 is supplied with the sampling difference ΔEL from the subtraction circuit 11, and when the sign of this sampling difference ΔEL is positive, it performs an addition operation, and when the sign of this sampling difference ΔEL is negative, it performs a subtraction operation. , a control circuit 23 controls the adder/subtractor 21. The adder/subtractor 31 is configured to perform either addition or subtraction operation based on the output of the control circuit 33. The control circuit 33 is supplied with the sampling difference ΔER from the subtraction circuit 12, and when the sign of this sampling difference ΔER is positive, it performs an addition operation, and when the sign of this sampling difference ΔER is negative, it performs a subtraction operation. Then, the control circuit 33 controls the adder/subtracter 31.

The adder/subtractor 41 is configured to perform either addition or subtraction operation based on the output of the control circuit 43. Control circuit 4
3 is supplied with the line difference ΔLU from the subtraction circuit 13,
The control circuit 43 controls the adder/subtractor 41 so that when the sign of this line difference ΔL[J is positive, an addition operation is performed, and when the sign of this line difference ΔLU is negative, a subtraction operation is performed.

The sampling difference ΔEL is supplied to a conversion circuit 24, converted to an absolute value, and supplied to one input of an adder 25 for integration. The output of the adder 25 is supplied to the register 26, the output of the register 26 is supplied to the other input of the adder 25, and the output of the register 26 is supplied with the sampling difference ΔEL.
The integrated value of the absolute value of is extracted. Sampling difference ΔE
R is supplied to a conversion circuit 34, converted to an absolute value, and supplied to one input of an adder 35 for integration.

The output of the adder 35 is supplied to the register 36, the output of the register 36 is supplied to the other input of the adder 35, and the integrated value of the absolute value of the sampling difference ΔER is taken out as the output of the register 36.

The line difference ΔL[J is supplied to a conversion circuit 44, converted to an absolute value, and supplied to one input of an adder 45 for integration. The output of the adder 45 is supplied to the register 46, the output of the register 46 is supplied to the other input of the adder 45, and the integrated value of the absolute value of the line difference ΔLU is taken out as the output of the register 46. The line difference ΔLD is the conversion circuit 5
4, converted to an absolute value, and supplied to one input of an adder 55 for integration. The output of the adder 55 is supplied to the register 56, and the output of the register 56 is supplied to the adder 5.
5, and the integrated value of the absolute value of the line difference ΔLD is taken out at the output of the register 56.

The horizontal motion vl is determined by the divider 27 or 37 by adding or subtracting the frame difference ΔF in the motion area and dividing it by the integrated value of the absolute value of the sampling difference ΔEL or ΔER. Further, the vertical motion v2 is obtained by adding or subtracting the frame difference ΔF in the motion area using the divider 47 or 57, and dividing it by the integrated value of the absolute value of the line difference ΔLU or ΔLD.

The motion output detected from the divider 27 using the sampling difference ΔEL on the left side is supplied to the selector 28 and the select signal generation circuit 29, and the motion output detected using the sampling difference ΔEL on the right side from the divider 37 is supplied to the selector 28 and the select signal generation circuit 29. selector 28
and is supplied to the select signal generation circuit 29. This selector 2B is controlled by a select signal from a select signal generating circuit 29. The output signal of the selector 2 is taken out to the output terminal 30 as a horizontal movement output vl. In this motion output v1, its sign bit indicates the direction of motion, and its value indicates the amount of motion.

The motion output detected from the divider 47 using the upper line difference ΔLU is sent to the selector 48 and the select signal generation circuit 4.
9, and the motion output detected from the divider 47 using the lower line difference ΔLD is supplied to the selector 48 and the select signal generation circuit 49. This selector 48 is
It is controlled by a select signal from a select signal generating circuit 49. The output signal of the selector 48 is taken out to the output terminal 50 as a vertical movement output v1.

In this motion output v1, its sign bit indicates the direction of motion, and its value indicates the amount of motion.

The select signal generation circuit 29 is configured to generate a select signal that selects the one with a larger absolute value among the two motion outputs supplied to the selector 28, or according to the sign of the two motion outputs supplied to the selector 28. (That is, when the sign is positive, select the motion output based on the left sampling difference ΔEL, and when the sign is 1, select the motion output based on the right sampling difference ΔEL.
1 construction[? Select the motion output that sticks to the ) Select motion output.

The select signal generating circuit 49 similarly generates a select signal for the selector 48.

As described above, by providing two independent motion detection systems in each of the horizontal and vertical directions and selecting the motion output of each system, the accuracy of motion detection can be improved. The reason is explained below.

When the three-pixel object shown by the broken line in Figure 3 moves to the right after one frame as shown by the solid line, the gradient method calculates the slope ΔEL of point b (current pixel) between (a - b). It is determined by the difference. That is, the left slope Δ of the current pixel
The rightward movement is calculated from the relationship with the frame difference ΔF using EL. As for the slope, the right side of the current pixel <b-c
Even if the difference ΔER between ) is used, the same amount of motion can be found because all pixels are used for integration.

However, in the case of extreme values, the accuracy changes significantly depending on whether ΔEL or ΔER is used as the slope.

When the three pixels including the extreme value indicated by the broken line in Fig. 4 move to the left by one pixel after one frame (movement amount VL)
Also, consider a case where these three pixels move one pixel to the right after one frame (movement amount vR). In this case, the frame difference ΔF is positive regardless of the direction of movement.

In the case of leftward movement, the left sampling difference ΔEL
By using , it is possible to accurately detect a movement to the left with high precision. However, using the sampling difference ΔER on the right,
Since this sampling difference ΔER is negative, it is erroneously determined as a rightward movement, and the detection accuracy is poor. On the other hand, in the case of rightward movement, if the right side sampling difference ΔER is used, it is accurately detected as a rightward movement with high accuracy. However, when the sampling difference ΔEL on the left side is used, since the sampling difference ΔEL is positive, it is erroneously determined as a leftward movement, and the detection accuracy is poor. In this way, the accuracy of motion detection is completely different depending on the direction of movement, depending on whether the left side ΔEL or the right side ΔER is used as the slope.

Regardless of whether the direction of the extreme value is upward or downward, ΔE
The relationship between L and ΔER and leftward movement and rightward movement is shown in the following table.

In the embodiment of the present invention described above, in the horizontal direction, ΔE
Two systems of motion detection using each of L and ΔEH are provided,
Since the correct motion output is selected by the selector 28,
Motion detection can be performed with high accuracy. That is, as can be seen from the table above, the motion output with the larger absolute value is selected, or when the sign of the motion output is positive, the motion output of the system using ΔEL is selected, and the motion output of the motion output is selected. When the sign is negative, the selector 28 selects the motion output based on the sign, so as to select the motion output of the system using ΔER.

Regarding the movement in the vertical direction, similarly to the case in the horizontal direction, the selector 48 selects the correct one of the two motion outputs using the line differences ΔLU and ΔLD as the inclinations.

Note that the present invention may be applied to one-dimensional motion detection that detects motion only in either the horizontal direction or the vertical direction. Further, the present invention can also be applied to a motion detection circuit that does not use all pixels but uses representative points for each predetermined number of pixels.

In addition, the control circuits 23, 33, 43.53 that control the adder/subtractor that integrates the frame difference are not limited to simple control based on the sign of the slope, and hold the previous state until the sign is reversed. It is also possible to have a configuration in which According to this configuration,
Even when the amount of motion is large or when a moving object has a steep slope, it is possible to prevent the accuracy of motion detection from decreasing.

[Effects of the Invention] According to the present invention, motion detection can be performed correctly even for extreme values that do not have stationarity, and therefore the accuracy of motion detection can be greatly improved.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an embodiment of this invention, FIGS. 2, 3, and 4 are route diagrams used to explain an embodiment of this invention, and FIG. 5 is a movement diagram using the gradient method. FIG. 6 is a block diagram of a conventional motion detection circuit, and FIG. 7 is a route diagram used to explain problems with the conventional motion detection circuit. 1: Digital television signal input terminal, 2゜3 line delay circuit, 4, 5, 6.8: Sample delay circuit,
7: Frame delay circuit, 10.11°12.13,14
: Subtraction circuit, 21, 31, 41° 51: Addition/fj, multiplier, 27. 37, 47, 57: Divider, 28. 48: Selector, 29. 49: Select signal generation circuit, 30, 50
: Output terminal.

Claims (1)

[Scope of Claims] Delay and calculation means for generating a frame difference between a current frame and a previous frame of a digital television signal; a delay and calculation means for generating a slope of the digital television signal; and a system for integrating the frame difference. an adder/subtracter constituting an integrating means for calculating the value; a means for dividing the output of the integrating means by the integrated value of the absolute value of the slope of the digital television signal; and addition or subtraction of the adder/subtracter. at least two motion detection circuits comprising means for controlling the state of the current pixel in relation to the polarity of the slope of the digital television signal; The other of the motion detection circuits is configured to detect the tilt between the current pixel and the next pixel using the delay and calculation means, and detects the tilt between the current pixel and the next pixel, and detects the tilt between the current pixel and the next pixel using the delay and calculation means. A motion detection circuit for a television signal, comprising means for selecting and outputting motion outputs of the two motion detection circuits based on the information.