JPH06153167A - Motion vector detection circuit - Google Patents

Abstract

PURPOSE:To reduce dispersion in motion vectors by comparing plural blocks of detected motion vectors so as to suppress production of an isolated motion vector and to smooth the motion vectors. CONSTITUTION:The circuit is provided with motion vector detection means 2-7 dividing a pattern minutely into blocks comprising prescribed picture element number and line number to detect a motion vector MV for each block, an isolation vector elimination means 11 comparing plural blocks of the detected motion vectors MV by the vector detection means 2-7 to eliminate the isolated motion vector MV in the motion vectors of plural blocks, and a filter means 12 smoothing the motion vector MV10 sent from the isolation vector elimination means 11. Then when a picture is divided minutely in the unit of blocks to detect the motion vector MV, plural blocks of the detected motion vector MV are compared to eliminate an isolated motion vector MV, resulting that the obtained motion vectors MV 10 are smoothed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a motion vector detection circuit, and can be applied to, for example, a device for detecting motion information (that is, motion vector) of an image in a television system conversion device using a digitized television signal.

[0002]

2. Description of the Related Art Conventionally, in a TV system conversion apparatus for converting a television system between an NTSC system and a PAL system using a digitized television signal, a motion vector is detected from the television signal to perform motion compensation. There are some that are designed to reduce the discontinuity of motion due to the conversion of the number of fields.

In detecting the motion vector, the television signal is subdivided into blocks of m pixels × n lines (m and n are integers), and then the motion vector is detected for each block. For example, a pattern matching method (JP-A-55-162683, JP-A-55-162684)
And the iterative gradient method (Japanese Patent Laid-Open No. 60-158786) are used.

In practice, as shown in FIG. 2, in the motion vector detection circuit 1 in the conventional TV system conversion device, the television signal is subdivided into blocks of 8 pixels × 8 lines, and the motion vector is calculated for each block. First, the luminance signal S1 of the current field and the luminance signal S2 of the previous field are input to the two-dimensional low-pass filters (LPF) 2 and 3, respectively.

The two-dimensional low-pass filters 2 and 3 remove noise and high-frequency components from the luminance signals S1 and S2 for one block respectively inputted, and the luminance signals S3 and S4 obtained as a result are respectively initialized. It is input to the displacement vector selection circuit 4 and the displacement vector detection circuit 5.

The initial displacement vector selection circuit 4 reads out several types of motion vectors MV0 already detected in a block in the vicinity of the detected block from the motion vector memory 6 configured to store the detected motion vector MV. , Pick out the best of these. Actually, the coordinates of the block are deviated by the amount of each motion vector MV0, the absolute value of the inter-field difference value is obtained, and the motion vector MV1 that gives the block with the smallest accumulated value in the block is determined to be optimum. It is designed to do.

The motion vector M selected in this way
V1 is input to the displacement vector detection circuit 5. In this displacement vector detection circuit 5, the initial displacement vector MV
After displacing the block coordinates by one minute, the motion vector MV2 is detected using the iterative gradient method.

The motion vector MV2 thus obtained and the initial displacement vector MV1 selected by the initial displacement vector selection circuit 4 are added by the adder circuit 7, and the resulting motion vector MV is output from the motion vector detection circuit 1. And is stored in the motion vector memory 6.

In the TV system converter, the motion vector MV thus detected is used as a vector for forming an interpolation signal for motion correction.

[0010]

However, in the motion vector detection circuit 1 having such a configuration, the motion vector MV
Is detected by using a block-unit iterative gradient method consisting of a predetermined number of pixels × lines, so that the detected motion vector MV may vary in block units.

In practice, when a motion vector MV is used to form an interpolated signal as in a TV format converter, the accuracy of the motion vector MV may greatly affect the image quality.
For example, as shown in FIG. 3A, in the image in which the pattern PTN of the vertical line moves in the direction of the arrow a in the screen DSP, when the detected motion vector MV is not uniform, As shown in (B), a vertical line pattern PT
Distortion in block units occurs in NX, and as a result, there is a problem that image quality is significantly deteriorated.

The present invention has been made in consideration of the above points, and proposes a motion vector detection circuit capable of reducing the variation of the motion vector when detecting the motion vector by subdividing an image into block units. It is a thing.

[0013]

In order to solve such a problem, in the present invention, a motion vector detecting means 2 for subdividing an image into blocks having a predetermined number of pixels and a number of lines and detecting a motion vector MV for each block. 3, 4,
5, 6, 7 and a plurality of blocks of the motion vector MV detected by the motion vector detecting means 2 to 7 are compared,
An isolated vector removing unit 11 that suppresses an isolated motion vector MV among the motion vectors MV for the plurality of blocks.
And a filter means 12 for smoothing the motion vector MV10 sent from the isolated vector removing means 11.

[0014]

When the image is subdivided into blocks and the motion vector MV is detected, a plurality of blocks of the detected motion vector MV are compared to suppress the isolated motion vector MV and the resulting motion is obtained. Vector MV1
By smoothing 0, the motion vector M
The variation of V11 can be reduced.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings.

In FIG. 1 in which parts corresponding to those in FIG. 2 are designated by the same reference numerals, reference numeral 10 generally indicates a motion vector detecting circuit according to the present invention, which is used in a TV system conversion device as in the conventional case. This motion vector detection circuit 10
In the conventional method, the television signal is subdivided into blocks of 8 pixels × 8 lines, and the motion vector is detected for each block.

Also in the motion vector detecting circuit 10, first, the luminance signal S1 of the current field and the luminance signal S2 of the previous field are input to the two-dimensional low pass filters (LPF) 3 and 4, respectively, to remove noise and increase the noise. The range component is removed, and the resulting luminance signal S
3 and S4 are input to the initial displacement vector selection circuit 4 and the displacement vector detection circuit 5.

The initial displacement vector selection circuit 4 reads several types of motion vectors MV0 already detected in the block near the detected block from the motion vector memory 6 as in the conventional case, and selects the optimum one among them. The motion vector MV1 is input to the displacement vector detection circuit 5. In the displacement vector detecting circuit 6, after the coordinates of the block are displaced by the initial displacement vector MV1,
The motion vector MV2 is detected using the iterative gradient method.

The motion vector MV2 and the initial displacement vector MV1 selected by the initial displacement vector selection circuit 5 are added by the addition circuit 7, and the resultant motion vector M is obtained.
In addition to V being stored in the motion vector memory 6,
In the case of this embodiment, it is input to the isolated point removing circuit 11.

The isolated point removing circuit 11 includes an isolated point removing IC integrated into a circuit by a so-called median filter. This median filter is designed so that signals can be taken out in descending order of signal level from a total of 9 samples of 3 × 3. For example, if the level is 4, then 9
It is possible to extract the signal of the fourth highest level from the samples.

In the case of this embodiment, for example, as shown in FIG. 4, the motion vectors MV of a total of 9 blocks B1 to B9 centering on the target block BOB (B5), 3 blocks in the horizontal direction and 3 blocks in the vertical direction, As a sample, for example, the level is set to 4, and a motion vector is taken out. In this way, the isolated point can be removed by suppressing the motion vector isolated from the periphery.

In practice, the isolated point removing circuit 11 is a combination of delay circuits 13 and 14 each having a delay time of one horizontal scanning period, as shown in FIG. It is configured so as to be taken into the isolated point removal IC 11A one by one. In this way, the motion vector MV10 from which the isolated points are removed is smoothed by the two-dimensional low-pass filter (LPF) 12, and the motion vector MV11
Is sent as.

This two-dimensional low-pass filter 12 is shown in FIG.
As shown in FIG. 5, a 5-tap FIR type digital circuit including delay circuits 15A to 15D having a delay time of one clock in the horizontal direction, integrators 16A to 16E for accumulating coefficients k1 to k5, and an adder 17 Filter and 1 vertically
Delay circuits 18A to 18 each having a delay time corresponding to the horizontal scanning period
An integrator 19 for integrating D and the coefficients k6 to k10
A to 19E and a 5-tap FIR digital filter composed of an adder 20.

In the above configuration, in the motion vector detecting circuit 10, the television signal is 8 pixels × 8.
It is subdivided into line blocks, and first, the luminance signal S1 of the current field and the luminance signal S of the previous field in block units.
2 is used to detect the motion vector MV by an iterative gradient method referencing the initial displacement vector.

Subsequently, with respect to this motion vector MV, the isolated points are suppressed and smoothed by referring to the motion vectors of the surrounding 3 × 3 blocks, and as a result, the motion vector MV obtained by the iterative gradient method is blocked. Even if there are variations in units, it is possible to effectively suppress them and detect a uniform motion vector MV11.

By transmitting the motion vector MV11 in which the variation in block units is suppressed in this way, even when the interpolation signal is formed using the motion vector MV as in the TV system converter, the motion vector MV is generated. It is possible to prevent the deterioration of the image quality due to the variation of each block unit.

According to the above configuration, when the image is subdivided into blocks and the motion vector is detected, a plurality of blocks of the detected motion vector are compared to suppress the isolated motion vector, and By smoothing the resulting motion vector, it is possible to realize a motion vector detection circuit that can reduce variations in the motion vector.

In the above embodiment, the isolated point removing circuit refers to the motion vectors of 3 × 3 blocks in the vertical and horizontal directions to remove the isolated points two-dimensionally. Alternatively, the isolated points may be removed by referring to a plurality of blocks, and instead of or in addition to this, the isolated points may be removed in the time axis direction. The same effect as can be achieved.

In this case, as shown in FIG. 7, in the isolated point removing circuit 21, the delay circuits 22 and 23 each having a delay time of one field are combined, and three blocks in the time axis direction are horizontally moved. Removal of isolated points by 3 blocks I
It is configured to be taken into C21A.

In the above embodiment, the case where the motion vector after isolated point removal is smoothed by a two-dimensional low-pass filter has been described. However, the present invention is not limited to this, and the characteristics are slightly deteriorated but the configuration is reduced. A simple one-dimensional low-pass filter may be used.

Further, in the above-described embodiment, the case where the motion vector is detected in block units by the iterative gradient method is described first, but instead of this, the motion vector is detected by another method such as the pattern matching method. Alternatively, when detecting the motion vector in block units, the same effect as that of the above-described embodiment can be obtained.

Further, in the above-described embodiment, the case where the present invention is applied to the one for detecting the motion vector for generating the internal signal in the TV system conversion device has been described, but the present invention is not limited to this and, for example, the motion vector is It is also suitable for wide application to the one for high-efficiency coding of an image using.

[0033]

As described above, according to the present invention, when an image is subdivided into blocks and a motion vector is detected, a plurality of blocks of the detected motion vector are compared to determine an isolated motion vector. By suppressing the motion vector and smoothing the resulting motion vector, it is possible to realize a motion vector detection circuit that can reduce the variation in the motion vector.

[Brief description of drawings]

FIG. 1 is a block diagram showing an embodiment of a motion vector detection circuit according to the present invention.

FIG. 2 is a block diagram showing a configuration of a conventional motion vector detection circuit.

FIG. 3 is a schematic diagram used for explaining deterioration of an image when motion vectors are not uniform.

FIG. 4 is a schematic diagram for explaining an isolated point removal method in the motion vector detection circuit of FIG.

5 is a block diagram showing an isolated point removal circuit in the motion vector detection circuit of FIG.

6 is a connection diagram showing a configuration of a two-dimensional low-pass filter in the motion vector detection circuit of FIG.

FIG. 7 is a block diagram showing another embodiment of the isolated point removing circuit.

[Explanation of symbols]

1, 10 ... Motion vector detection circuit 2, 3, 12 ...
Two-dimensional low-pass filter, 4 ... initial deviation vector detection circuit, 5 ... deviation vector detection circuit, 6 ... motion vector memory, 7 ... adder, 11 ... isolated point removal circuit.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yoshihiro Yamamoto 2-2-1, Jinnan, Shibuya-ku, Tokyo Inside the Japan Broadcasting Corporation

Claims (1)

[Claims] 1. A motion vector detecting means for subdividing an image into blocks each having a predetermined number of pixels and a number of lines and detecting a motion vector for each block, and a plurality of blocks of the motion vector detected by the motion vector detecting means. And an isolated vector removing unit that suppresses the isolated motion vector among the motion vectors of the plurality of blocks, and a filter unit that smoothes the motion vector sent from the isolated vector removing unit. A motion vector detection circuit, which is characterized by