JP2874654B2 - Motion detection circuit - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a motion detection circuit, and more particularly to a motion detection circuit for detecting a motion vector for compressing a video signal.

When transmitting or recording a video signal, a method of compressing the video signal is used to reduce the signal band or the amount of a recording medium. Here, when compressing a video signal, it is effective to perform motion compensation.

The motion compensation is performed by detecting a motion in a search range of N × M (N ≧ n, M ≧ m) in combination with a motion detection circuit in a range of n × m pixels.

In such a motion detection circuit, when the search range is widened, erroneous detection of a motion vector is affected by the influence of image noise and errors in A / D conversion and Y / C separation. It is necessary to prevent this from happening.

[0005]

2. Description of the Related Art The compression of a video signal uses a fact that a video signal has a strong correlation with a preceding or succeeding screen (frame or field). This is done by giving a sign. The motion detection circuit is used for detecting a motion vector for compressing a video signal.

When performing motion compensation in order to compress a video signal, a motion detection circuit in the range of n × m is combined to perform motion compensation in a search range of N × M (N ≧ n, M ≧ m). Perform motion detection.

Conventionally, a motion detection circuit is provided in which a plurality of processors for processing are provided in parallel to increase the processing speed and obtain a correct motion vector (see Japanese Patent Application Laid-Open No. 05-037922).
And a method of detecting a vector having the least correction error as a motion vector as a result of detecting an approximate vector of a motion vector and performing a plurality of motion corrections using the approximate vector and peripheral vectors (Japanese Patent Laid-Open No. 02-206985). (See FIG. 1), and erroneous detection of motion detection is prevented by changing the non-linear processing according to the ratio between the moving part and the stationary part of the entire screen.
No. 67) is known.

FIG. 4 is a block diagram showing an example of a conventional motion detection circuit. This circuit, the differential circuit 11 ₁ to 11
_k , a comparison circuit 12, and a selector 13.
The differential circuit 11 ₁ to 11 _k, commonly, the reference data (vertical m × horizontal n), the search range (vertical M ×, each share
The search data of the horizontal N, where M> m, N> n) is input, and the difference data that minimizes the difference result with the reference data within the search range and the data of the motion vector at that time are output.

In the comparison circuit 12, the difference circuits 11 _{1 to} 11
_The difference data output from _k is compared to generate a select signal indicating a difference circuit that minimizes the difference data. Moreover, the vector data outputted from the difference circuit 11 ₁ to 11 _k are inputted to the selector 13. The selector 13 selects the data of the minimum motion vector in the entire search range by the select signal from the comparison circuit 2. In this case, the wider the search range, the quicker the movement on the screen can be followed, and efficient video signal compression becomes possible.

[0010]

In the conventional motion detection circuit shown in FIG. 4, as the search range is widened, noise included in an analog video signal or a luminance signal from an A / D conversion or a composite video signal is reduced. And Y separating carrier color signal
Reference data and search data are affected by an error generated at the time of / C separation, and as a result, the probability of outputting difference data and an erroneous value of a motion vector increases. And when the vector data becomes larger than it actually is,
There is a problem that the compression efficiency of the video signal is deteriorated and the image quality is deteriorated.

As described above, in the conventional motion detection circuit, the reason that the probability that an erroneous value of the motion vector data is output becomes high is that the probability of false detection of the motion vector becomes high because the search range is large. .

[0012]

SUMMARY OF THE INVENTION The object of the present invention is to improve the disadvantages of the prior art, and in particular, to suppress the occurrence of a larger than actual motion vector caused by erroneous detection by changing the search range according to the motion of the video. It is an object of the present invention to provide a motion detection circuit capable of effectively preventing a reduction in the function of signal compression efficiency.

[0013]

According to the motion detecting circuit of the present invention, the search range changes according to the motion of the video. That is, motion detection is performed from a low-resolution video in advance, and a search range for normal-resolution motion detection is determined from the result. For images with small horizontal or vertical motion, narrowing the search range in the horizontal or vertical direction
It is possible to reduce erroneous detection of a motion vector at a normal resolution.

More specifically, a plurality of sampling circuits thin out input reference data and search data, thereby outputting low-resolution video data. Then, motion detection is performed on the low-resolution video data, difference data as a difference result and motion vector data are output, and the first comparison circuit compares the difference data output from the plurality of first difference circuits. Then, a difference circuit with the smallest difference data is selected, and a motion vector corresponding to the selected difference circuit is output. The control circuit performs a motion based on the motion vector output from the first comparison circuit. Output a control signal for limiting the search range in the direction in which is smaller, and the plurality of second difference circuits output images of normal resolution within the search range limited by the control signal. For data by performing motion detection, and outputs the difference data and the motion vector of the difference result,
The second comparison circuit compares the difference data output from the plurality of second difference circuits, and outputs a select signal for selecting a difference circuit with the smallest difference data. A motion vector is output from the selected difference circuit.

In the motion detecting circuit according to the present invention, erroneous detection of a motion vector is reduced in this manner, so that it is possible to prevent the compression efficiency of a video signal from being lowered.

Hereinafter, specific means for solving the problems of the present invention will be described.

[0017] (1) the motion detection circuit of the present invention includes a plurality of sampling circuits 1 ₀ to 1 _k to obtain a low-resolution image data by sampling by decimating the respective reference data and a plurality of search data to be input, this low-resolution image data by performing motion detection within a predetermined search range, a plurality of first differential circuit 2 ₁ to 2 _k for outputting the difference data and the motion vector data of the difference results,
The outputs a motion vector corresponding to the selected difference circuit along with differential data by comparing the differential data output from a plurality of first differential circuit 2 ₁ to 2 _k selects the differential circuit becomes minimum 1 Comparison circuit 3 and first comparison circuit 3
A control circuit 4 for outputting a control signal for restricting a search range in a direction in which motion is small based on a motion vector output from the CPU, and performing motion detection on video data having a normal resolution within the search range limited by the control signal the difference result and a plurality of second differential circuit 5 ₁ to 5 _k to output the difference data and the motion vectors of a plurality of the second differential circuit 5 ₁
, A second comparison circuit 6 that outputs a select signal for selecting a difference circuit that minimizes the difference data by comparing the difference data output from ５5 _k, and a motion vector from the difference circuit selected according to the select signal. And an output selector 7.

(2) The motion detection circuit of the present invention performs motion detection within a predetermined search range for input reference data and a plurality of search data, and outputs difference data of a difference result and motion vector data. A plurality of difference circuits 11 ₁ to
And 11 _k, compared to select the differential circuit the difference data is minimized by comparing the differential data output from the plurality of difference circuits 11 ₁ to 11 _k, outputs a select signal corresponding to the selected difference circuit A circuit 12, a selector 13 for outputting a motion vector from a differential circuit selected according to the select signal, and a correlation between the previously output motion vector and the currently output motion vector, and of a control circuit 14 for outputting a control signal for limiting the search range, the differential circuit 11 ₁ to 11 _k performs motion detection within a limited search range in accordance with the control signal.

(3) In the case of the above (1) or (2), the present invention
The, the control circuit, the search range of the motion detection in the first differential circuit 2 ₁ to 2 _k or differential circuit 11 ₁ to 11 _k,
If a horizontal positive motion vector equal to or more than a predetermined value continues, the horizontal search range is narrowed only in the positive direction.If a horizontal negative motion vector equal to or more than the predetermined value continues, the horizontal search range is reduced. When narrowing only in the negative direction and a horizontal motion vector having an absolute value equal to or less than a predetermined value continues, the horizontal search range is narrowed to a positive or negative predetermined value, and a vertical positive motion vector equal to or more than a predetermined value follows. In this case, the vertical search range is narrowed only in the positive direction.If a vertical negative motion vector that is equal to or more than a predetermined value continues, the vertical search range is narrowed only in the negative direction. If the vertical motion vector continues, the function is to narrow the vertical search range to a predetermined positive or negative value.

[0020]

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

[0021]

First Embodiment FIG. 1 shows a first embodiment of the present invention.

[0022] In FIG. 1, a sampling circuit 1 ₀
Is a circuit for thinning out reference data, and a sampling circuit 1
_{1 ~1 k} is a circuit for decimating the search data. Also,
First difference circuit 2 ₁ to 2 _k from the reference data and the search data thinned, it has a function of outputting the difference data and the motion vector. In the first comparator circuit 3, among the differential data output from the first differential circuit 2 ₁ to 2 _k, looking for something to be minimized, and outputs the motion vector from the origin of the reference data at that time. Further, the control circuit 4 outputs a predetermined control signal for changing the search range for the second differential circuit 5 ₁ to 5 _k based on the motion vector from the first comparator circuit 3.

[0023] within a limited search range, the second differential circuit 5 ₁ to 5 _k is the difference result between the reference data and the search data is not thinned out is minimized, and outputs the difference data and the motion vector. The second comparison circuit 6 includes a second difference circuit 5
Comparing the difference data from ₁ to 5 _k, identifies a difference circuit which differential data is minimized, and outputs a select signal.
The selector 7 inputs a motion vector from the second difference circuit 6 and outputs a motion vector with a minimum difference result according to a select signal from the second comparison circuit.

The operation of the first embodiment shown in FIG.
A description will be given based on FIG. Sampling circuit _{1 1 ~1 k}
As shown in FIG. 2, reference data of a reference block having m pixels in the vertical direction and n pixels in the horizontal direction is commonly input, and a search range of N pixels in the vertical direction and M pixels in the horizontal direction is defined. Then, search data in a range divided into areas that can be searched by each difference circuit is input, and an operation of thinning out the data is performed.

The first difference circuits 2 ₁ to 2 _k are used to calculate the difference data that is the smallest between the reference data and the search data that have been decimated in the area that can be searched by each difference circuit, and the difference data of each difference circuit at that time. Detect the motion vector from the origin. In the first comparator circuit 3, looking for something to be minimized among the differential data output from the first differential circuit 2 ₁ to 2 _k, and outputs the motion vector from the origin of the reference data at that time.

The control circuit 4 narrows the search range in the vertical direction when the motion vector is large in the horizontal direction based on the motion vector from the first comparison circuit, and narrows the search range in the horizontal direction when the motion vector is large in the vertical direction. A control signal for narrowing the search range is output. The search range is changed by this signal. Within a limited search range, the second differential circuit 5 ₁ to 5 _k compares the reference data and the search data is not thinned out, the difference data difference result is minimized, the movement from the origin of the difference circuit at that time Output a vector.

The second comparison circuit 6 includes a second difference circuit 5 ₁
Comparing the difference data from to 5 _k, the difference data to output a select signal identifying the difference circuit to be minimized. The selector 7 inputs a motion vector from the second difference circuit, and determines a motion vector that minimizes the difference result in a search range of N pixels in the vertical direction and M pixels in the horizontal direction by a select signal from the second comparison circuit. Output.

As a result, the minimum vector data of the motion vector is always detected, which causes the disadvantage of the conventional example that the vector data becomes larger than the actual one and the compression efficiency of the video signal is deteriorated. It can be effectively improved and, therefore, its use can effectively prevent image quality degradation.

[0029]

Second Embodiment FIG. 3 shows a second embodiment of the present invention. Here, the same components as those in the above-described conventional example are denoted by the same reference numerals.

In FIG. 3, a control circuit 14 is provided at the output stage of the selector 13. In this control circuit, a motion vector is input, and a motion vector input before that is held, these motion vectors are compared, and the correlation of the motion vector over the entire screen is used. A control signal for limiting the search range in the horizontal or vertical direction is output.

The control circuit 14 narrows the search range in the horizontal direction only in the positive direction when the horizontal motion vector continues for more than +32, for example. Functions to narrow the range only in the negative direction. Further, the control circuit 14 functions to narrow the horizontal search range to ± 32 when the motion vector continues to be, for example, 16 or less in absolute value. On the other hand, if the vertical motion vector continues to be, for example, +16 or more, the vertical search range is narrowed only in the positive direction. If the vertical motion vector continues to be -16 or less, the vertical search range is narrowed only to the negative direction. To work. If the absolute value continues to be 8 or less,
It functions to narrow the vertical search range to ± 16.

For this reason, since the control circuit 14 functions to narrow the search range in any case, it is possible to almost completely improve the disadvantage of the conventional example that erroneously outputs large vector data.

[0033]

Next, an embodiment of the present invention will be described with reference to the drawings. In the first embodiment shown in FIG. 1, the sampling circuit _{1 1 ~1 k (k = 32} ), as shown in Figure 2, in the longitudinal direction m = 8 pixels, lateral n = 16 pixels of the reference block (Reference data) are commonly input. Further, the search range (vertical direction ± 32, horizontal direction ± 64) in which the vertical direction M = 64 pixels and the horizontal direction N = 128 pixels is divided into an area (16 pixels × 16 pixels) where each difference circuit can search. Search data is input. Then, an operation of thinning out the data in, for example, half in both the vertical and horizontal directions is performed.

First difference circuits 2 ₁ to 2 _k (k = 16)
Calculates the difference between the reference data and the search data within the area that can be searched by each difference circuit, and detects the difference data that minimizes the result and the motion vector from the origin of each difference circuit at that time. The first comparison circuit 3 compares the difference data output from the first difference circuits 2 ₁ to 2 _k (k = 16) to find the smallest one, and finds the origin of the entire search range in the difference circuit. Output the motion vector from.

The control circuit 4 outputs a control signal for limiting the search range based on the motion vector from the first comparison circuit 3. When the horizontal motion vector is 32 or more, the horizontal search range is narrowed only in the positive direction, and -3
If the number is 2 or less, the horizontal search range is narrowed only in the negative direction. If the absolute value is 16 or less, the search range in the horizontal direction is narrowed to ± 32. On the other hand, when the vertical motion vector is 16 or more, the vertical search range is narrowed only in the positive direction, and when it is -16 or less, the vertical search range is narrowed only in the negative direction. If the absolute value is 8 or less, the search range in the vertical direction is narrowed to ± 16.

[0036] within a limited search range in this way, the second differential circuit _{5 1 ~5 k (k = 16} ) compares the reference data and the search data is not thinned, difference data difference result is minimized Then, a motion vector from the origin of each difference circuit at that time is output. Second comparison circuit 6 compares the difference data from the second differential circuit _{5 1 ~ k (k = 16} ), and outputs a select signal identifying the difference circuit the difference data is minimal. The selector 7 inputs a motion vector from the second difference circuit, and determines a motion vector with a minimum difference result in a search range of 64 pixels in the vertical direction and 128 pixels in the horizontal direction by a select signal from the second comparison circuit. Can be output.

[0037]

As described above, according to the present invention,
The control circuit sets the search range for motion detection in the difference circuit.
Since it is changed so that it becomes narrower according to the motion of the video, it is possible to reliably reduce the occurrence of a motion vector larger than the actual one, and therefore, it is possible to effectively prevent erroneous detection of the motion vector. Then, it is possible to provide an unprecedented excellent motion detection circuit that can reliably prevent a decrease in the compression efficiency of the video signal.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a first embodiment of the present invention.

FIG. 2 is a diagram for explaining the operation of FIG. 1;

FIG. 3 is a block diagram showing a second embodiment of the present invention.

FIG. 4 is a block diagram illustrating an example of a conventional motion detection circuit.

[Explanation of symbols]

 10-1k sampling circuit 21-2k first difference circuit 3 first comparison circuit 4,14 control circuit 51-5k second difference circuit 6 second comparison circuit 7,13 selector 111-11k difference circuit 12 comparison circuit

Claims (2)

(57) [Claims] 1. A plurality of sampling circuits for obtaining low-resolution video data by thinning out and sampling a plurality of input search data, respectively, and a plurality of sampling circuits provided corresponding to the respective sampling circuits. A plurality of first difference circuits for performing motion detection within a predetermined search range based on the output and separately input sampled reference data and outputting difference data of the difference result and motion vector data; First
And a first comparison circuit that compares the difference data output from the difference circuits to select a difference circuit with the smallest difference data and outputs a motion vector corresponding to the selected difference circuit. A control circuit for outputting a control signal for limiting a search range in a direction in which motion is small based on the motion vector output from the comparison circuit, and performing motion detection at a normal resolution within the limited search range to obtain the difference result. A plurality of second difference circuits for outputting the difference data and the motion vector, and comparing the difference data output from the plurality of second difference circuits to obtain a difference circuit that minimizes the difference data. A second comparison circuit that outputs a select signal to be selected; and a selector that outputs a motion vector from a difference circuit selected in accordance with the select signal. Next to the search range of the differential circuits in definitive motion detection, if the positive motion vector of lateral or greater than a predetermined value continues to function to narrow the lateral search range only in the positive direction, the predetermined value or more When a negative motion vector in the direction continues, the horizontal search range is narrowed only in the negative direction. When a horizontal motion vector with an absolute value less than a predetermined value continues, the horizontal search range is reduced. When a vertical positive motion vector equal to or more than a predetermined value continues while narrowing to a positive / negative predetermined value, the vertical search range functions to narrow only in the positive direction, and a vertical negative motion vector equal to or more than a predetermined value. In the case where continues, the vertical search range is narrowed only in the negative direction, and if the vertical motion vector having an absolute value equal to or less than a predetermined value continues, the vertical search range is narrowed to a predetermined positive or negative value. To work A motion detection circuit characterized by the following. 2. A plurality of difference circuits for performing motion detection within a predetermined search range based on input reference data and a plurality of search data and outputting difference data and motion vector data of a difference result; A comparison circuit that compares the difference data output from the difference circuits to select a difference circuit with the smallest difference data and outputs a select signal corresponding to the selected difference circuit, and a select output from the comparison circuit A selector that functions in accordance with the signal and outputs a motion vector from a selected difference circuit. The selector retains the previously output motion vector and detects a correlation with the currently output motion vector to detect the motion. A control circuit for outputting a control signal for limiting a search range in a direction in which the difference is small, wherein each of the difference circuits includes a control signal from the control circuit. It operated in accordance with, comprising a motion detection function for motion detection within a limited search range, wherein the control circuit includes a search range of the motion detection in the previous SL differential circuit, a predetermined value or more lateral positive motion It works to narrow the horizontal search range only in the positive direction when a vector continues, and narrows the horizontal search range only in the negative direction when a horizontal negative motion vector equal to or more than a predetermined value continues. When a horizontal motion vector having an absolute value equal to or smaller than a predetermined value continues, the horizontal search range is narrowed to a predetermined positive or negative value, and when a vertical positive motion vector equal to or larger than the predetermined value continues, Functions to narrow the search range in the vertical direction only in the positive direction, and functions to narrow the search range in the vertical direction only in the negative direction when a negative vertical motion vector of a predetermined value or more continues. Absolute value of Motion detecting circuit, characterized in that function to narrow the vertical search range to a predetermined value of the positive and negative in the case where the vertical direction of the motion vector is followed.