JPH02274083A - Dynamic vector detecting device - Google Patents

Abstract

PURPOSE:To detect an exact dynamic vector by detecting the dynamic vector by block matching with respect to an image signal obtained by cutting a spatial high frequency component of a dynamic image signal. CONSTITUTION:A dynamic image signal (a) is cut as to its spatial high frequency component by a two-dimensional low-pass filter circuit 2. Subsequently, in a dynamic vector detecting circuit 3, an evaluation function is calculated with respect to thinned-out picture elements (for instance, 32 picture elements by thinning 1/2) in an MC block at every MC block. By what is called block matching, a dynamic vector is detected. In this case, although the dynamic vector detection by block matching is executed with respect to the thinned-out picture elements in the MC block, the high frequency component is cut in advance by the two-dimensional low-pass filter circuit 2 with respect to the image signal before thinning out the picture elements. In such a way, this device is not influenced by a turn-back noise by the high frequency component caused by thinning-out, uniform dynamic vectors can be obtained in each block, and an exact dynamic vector can be detected.

Description

DETAILED DESCRIPTION OF THE INVENTION [Industrial Application Field] The present invention relates to a coding device for highly efficient coding of a moving image signal using motion information, and particularly to a device for detecting motion information of a moving image signal. It is.

[Conventional technology]

Conventionally, in a device that detects a motion vector, which is a vector signal representing motion information of a video signal, the video signal is divided into blocks of a predetermined size (motion vector detection block, hereinafter referred to as rMC block), which is a unit of motion vector detection. ), and in each MC block, M
Pixels in the C block have been thinned out, and motion vectors have been detected by block matching using the thinned out pixels.

In FIG. 7, a moving image signal a divided into MC blocks is input to a moving vector detection circuit l via an input terminal T1. The motion vector detection circuit 1 thins out pixels in the MC block as shown by a motion vector detection block 1a, and detects a motion vector b by block matching using the thinned out pixels. In FIG. 7, T2 is an output terminal to which the detected motion vector b is output.

[Problem to be solved by the invention]

In the conventional motion vector detection device described above, the motion vector is detected by block masoching using thinned pixels in the MC block, so the scale of the hardware can be reduced. Since the MC block is thinned out, some of the information in the MC block is lost, making it difficult to detect an accurate motion vector signal.

[Means to solve the problem]

In order to eliminate such drawbacks, the motion vector detection device according to the present invention includes a low-pass filter circuit and a motion vector detection circuit, and the low-pass filter circuit suppresses spatial high-frequency components of the video signal. The motion vector detection circuit uses the motion image signal output from the low-pass filter circuit to
This method evaluates some pixels within a block and detects a motion vector by block matching.

[Effect]

In the motion vector detection device according to the present invention, an accurate motion vector can be detected even if block matching is performed using some thinned out pixels in the MC block.

〔Example〕

FIG. 1 is an explanatory diagram illustrating the present invention in detail. In the figure, TI is an input terminal, 2 is a two-dimensional low-pass filter circuit, 3 is a motion vector detection circuit, and 2a and 3a are M
C block, T2 is an output terminal.

In Figure 1, the size of MC blocks 2a and 3a is 8x8 (
64 pixels) (3×3 is omitted in the figure). First, the two-dimensional low-pass filter circuit 2 spatially cuts high-frequency components of the moving image signal a. Then, in the motion vector detection circuit 3, for each MC block, the thinned pixels in the MC block (for example, 32
Motion vector detection is performed by so-called flow matching, which calculates an evaluation function for each pixel (which becomes a pixel).

Here, motion vector detection will be performed by block matching on the thinned out pixels in the MC block, but the pixels are thinned out (previous image signals are processed using a two-dimensional low-pass filter). 2, the high-frequency components are captured, so there is no effect of aliasing noise in the high-frequency components due to thinning, and uniform motion vectors are obtained in each block, allowing accurate motion vector detection. becomes possible.

Next, an embodiment of the present invention will be described using FIGS. 2 to 6.

FIG. 2 is a block system diagram showing an embodiment of a motion vector detection device according to the present invention.

The apparatus shown in FIG. 2 manually inputs a moving image signal a from an input terminal T1 and obtains a moving vector b from an output terminal T2.

In the figure, a two-dimensional low-pass filter circuit 2 is a two-dimensional filter circuit that removes high frequency components in the spatial frequency domain. A simple example of this two-dimensional filter circuit is one expressed by the transfer function shown in the following equation.

Z-1・W-'(2+Z'+Z-')(2+W'+W
-')/4 where Z is a horizontal delay operator and W is a vertical delay operator.

The frame memory 4 is a memory circuit that delays the output image signal C of the two-dimensional low-pass filter circuit 2 by one frame. The motion vector detection circuit 3 is a circuit that receives an image signal C (current frame signal) and an image signal d (previous frame signal) and outputs a motion vector b. In this motion vector detection circuit 3, the difference between the current frame signal C and the previous frame signal d is calculated for the thinned out pixels for each MC block of a certain predetermined size (for example, 8 pixels x 8 lines). is calculated, and the sum of absolute values of these differences is calculated.

This calculation is performed for each trial vector, and the trial vector with the smallest value is determined to be the optimal motion vector and is outputted from the output terminal T2 as the motion vector b.

The details of the concrete implementation method of this motion vector detection circuit 4 can be found in the literature (Furuga, Hirano, Ohki, Iinuma, "Motion Compensated Interframe Coding of Conference Television Signals", IEICE Technical Report Gakoken Fund, ■
ε81-54, July, 1981).

Next, using FIGS. 3 to 6, it will be explained that an accurate motion vector can be detected with this apparatus.

FIG. 3(b) shows the pixel arrangement of the image signal a in FIG. 2 and its spatial frequency spectrum. Here, although the image signal is not band-limited as shown in FIG. 3(al), for convenience of description, it is assumed that the band is limited in a square shape as shown in FIG. 3(fbl).

4(al) and (b) show the pixel arrangement of the image signal C, which is the output of the two-dimensional low-pass filter circuit 2, and its spatial frequency spectrum. As can be seen from the dimensional spatial frequency spectrum, the high spatial frequency components of the image signal are cut.

FIG. 5(a) shows the pixel arrangement used in the motion vector detection circuit 3 to calculate the evaluation function within the MC block. As shown in FIG. 5 (al), in this embodiment, when detecting a motion vector, the pixels in the MC block are line-offseted and the pixels thinned out to 1/2 are subjected to block matching.

That is, the sum of absolute values of differences between frames is calculated for pixels thinned out within this MC block to obtain a motion vector. At this time, as shown in Figure 6, by thinning out pixels, , the high-frequency components are affected by aliasing noise. However, as can be seen from the spatial frequency spectrum shown in FIG. 5(b), the two-dimensional low-pass filter circuit 2 spatially cants the high frequency components, so it is not affected by this aliasing noise 5. , uniform motion vectors can be detected in each MC block.

〔Effect of the invention〕

As explained above, the present invention can remove the influence of aliasing noise in high-frequency components by performing motion vector detection by block matching on an image signal in which the spatial high-frequency components of a moving image signal are canted. , M.C.
Even if block matching is performed using some thinned out pixels within a block, an accurate motion vector can be detected.

[Brief explanation of drawings]

FIG. 1 is an explanatory diagram for explaining the present invention in detail, FIG. 2 is a block diagram showing an embodiment of the motion vector detection device according to the present invention, and FIGS. 3 to 5 are pixel arrangements of MC blocks. FIG. 6 is an explanatory diagram showing the conventional spatial frequency spectrum. FIG. 7 is an explanatory diagram showing the spatial frequency spectrum.
The figure is a block system diagram showing a conventional motion vector detection device. 2... Two-dimensional low-pass filter circuit, 3... Motion vector detection circuit, 4... Frame memory, T1... Input terminal, T2... Output terminal. Figure 1 Patent applicant NEC Corporation

Claims (1)

[Claims] A motion vector detection device for detecting a vector signal representing motion information of a video signal, comprising a low-pass filter circuit and a motion vector detection circuit, the low-pass filter circuit detecting spatial high-frequency information of the video signal. The motion vector detection circuit evaluates some pixels in the block using the video signal output from the low-pass filter circuit, and detects the motion vector by block matching. A motion vector detection device characterized by: