JPH08336141A - Movement vector detection device - Google Patents

Abstract

PURPOSE: To provide improved device in exactly specifying a motion vector, while using a median filtering technique. CONSTITUTION: An input digital video signal is inputted to a frame memory 100 and a method estimator 200, and frame data from the memory 100 are inputted to the estimator 200 as well. The estimator 200 specifies motion vectors to respective pixels included in a current frame, by using a block matching algorithm and supplies them to a median frame 300. The filter 300 averages the motion vectors by multiplying a predetermined filter coefficient, allocates, the average value as a filtered motion vector and supplies it to a majority detector 400. The detector 400 receives the filtered motion vector and specifies 1st and 2nd main vectors by counting the number of motion vectors having the equal value.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an improved apparatus for detecting motion vectors, and more particularly to an improved apparatus for detecting motion vectors using median filtering techniques.

[0002]

In various electronic / electrical applications, such as high definition television and video telephone systems, it is essential that video signals be transmitted in digital form. When the video signal is represented in digital form, a significant amount of digital data can be generated. However, since the available frequency bandwidth of a normal transmission channel is limited, the amount of transmission data must be compressed or reduced in order to transmit a video signal through the channel. Among various video compression techniques, motion-compensated interframe coding technique is known as one of the effective compression techniques, which is used for compressing a signal between two adjacent video frames. The temporal redundancy of the video signal is used.

In the motion compensation interframe coding technique,
Current frame data is predicted from the previous frame based on an estimate of motion between the current frame and the previous frame. Such estimated motion can be explained by a two-dimensional motion vector that represents the displacement of pixels between the previous frame and the current frame. One of the motion vector estimation techniques proposed in the art is the block matching algorithm (J.
R. "Displacement by Jain et al.
Measurement and Its Appli
situations in Interframe Ima
ge Coding ", IEEE Transacti
ons on Communications COM
-29, No. 12 (December 1981)).

According to the block matching algorithm, the current frame is divided into a plurality of search blocks of the same size. The search block size is typically in the range between 8x8 pixels and 32x32 pixels. In order to specify the motion vector to the search block in the current frame, similarity calculation is performed between the search block in the current frame and a plurality of candidate blocks having the same size as this search block. This candidate block is generally included in the search area in the previous frame, and the size of this search area is larger than that of the search block. Mean absolute error
solution error or mean square error (me
An error function, such as an square error, is used to measure the similarity between the search block of the current frame and each candidate block in the search region. By definition, the motion vector represents the displacement between the search block and the best matching block, ie the candidate block that yields the smallest "error" or difference.

In such motion estimation, it is preferable and convenient to search for only one minimum mean absolute error over the entire search area corresponding to the search block. However, there may be multiple occurrences of equivalent minimum differences between block matches. In this case, MPEG proposes to discard all equivalent minimum errors, even though they are not shown as standardization, except for their highest priority minimum error (ie the first error found). It was Therefore, with such a technique, it is difficult to accurately detect a motion vector between a search block and its corresponding search area.

[0006]

SUMMARY OF THE INVENTION Therefore, the main object of the present invention is to perform median filtering.
ltering) technique to provide an improved apparatus for accurately identifying motion vectors.

[0007]

In order to achieve the above object, according to the present invention, a motion vector used in a motion compensated video coder and representing a displacement between a current frame and a previous frame of a video signal is calculated. A motion vector, which is specific, the current frame being divided into a plurality of handling blocks having P × Q pixels, the P × Q pixels comprising a center pixel representing a pixel located at the center of each handling block. A motion vector (P and Q is an integer greater than 2) for providing a motion vector to each pixel of the current frame, the motion vector comprising a center motion vector for each center pixel. Providing means for generating a filtered motion vector for pixels included in the current frame by median filtering the motion vector. Of the handling blocks by counting the number of filtered motion vectors having the same value between the ian filtering means and the filtered motion vectors corresponding to the pixels included in each handling block. A sorting means for selecting the first main vector having the largest number and the second main vector having the second largest number, and an error function for each handling block based on the first and second main vectors. Selecting means for calculating and selecting, as an output motion vector, a vector that produces a minimum error function among the first and second main vectors; the center motion vector, the first main vector and the first motion vector for each handling block; Control signal generating means for generating a control signal based on the counted number of main vectors; In response to the control signal comprises and a switching means for selecting a motion vector from the center motion vector and the output motion vectors for each handling block.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The motion vector detecting device of the present invention will be described in more detail below with reference to the accompanying drawings.

Referring to FIGS. 1 and 2, there is shown a block diagram of an apparatus according to the present invention for identifying a motion vector for each handling block contained in the current frame, each handling block being P × Q pixels (for example, 5
× 5 pixels).

The current frame data is used as an input digital video signal for the frame memory 100 and the motion estimator 20.
Input to 0. Previous frame data from the frame memory 100 is also input to the motion estimator 200.

The motion vector estimator 200 operates to identify the motion vector for each pixel contained in the current frame using a block matching algorithm. In particular, in order to specify a motion vector to a pixel (target pixel) included in the current frame, the target pixel is used as a reference to match the center of the search block, and the target pixel is used as M × N pixels, for example, 3 pixels. A search block of × 3 pixels is set. This search block is compared on a block-by-block basis with all candidate blocks contained in the corresponding search area in the previous frame using a block matching algorithm known to those skilled in the art. The displacement for the search block obtained using the block matching algorithm is assigned as the motion vector of the target pixel. Each motion vector corresponding to each pixel included in the current frame is supplied from the motion estimator 200 to the median filter 300 for further processing. The motion vector estimator 200 also supplies the motion vector of the center pixel to the comparator 510 according to the present invention. Each center pixel is a pixel located at the center of each handling block if P and Q are odd integers, and if any one of P or Q is an even integer, then each center pixel of each handling block One of the two pixels located in the center,
If P and Q are even, it is one of the four pixels located at the center. That is, the center pixel is one of the maximum four pixels located at the center of the handling block.

In median filter 300, median filtering operations are performed on each motion vector from motion estimator 200 using the appropriate filtering window. In the preferred embodiment of the present invention, the median filter 300 has a cross-shaped filtering window as shown in FIG. 3, and predetermined filter coefficients for pixels A to E included in the cross-shaped filtering window. By averaging the motion vectors for them and assigning the averaged value as the filtered motion vector for pixel A. Each filtered motion vector corresponding to each pixel included in each handling block of the current frame is calculated by the median filter 30.
From 0 to the majority detector 400.

The majority detector 400 is, for example,
A filtered motion vector corresponding to each pixel included in each handling block of 5 × 5 pixels is received and the number of filtered motion vectors having an equal value is counted to determine the first and second main Identify the vector. The first and second main vectors respectively correspond to the vector having the largest number and the vector having the second largest number among the filtered motion vectors. The first and second main vectors are output through the output terminals 400A and 400C of the majority detector 400, respectively.
Also, the majority detector 400 has its output terminal 40.
Output the number of the first main vector through 0B.

The comparator 515 is a majority detector 4.
The number of first main vectors from 00 is compared with a predetermined value (for example, 13), and if the number of the compared first main vectors is 13 or more, a logical "high" signal is output. If so, it sends a logical "low" signal to AND gate 520. The comparator 510 compares the first main vector input from the output terminal 400A with the motion vector of the center pixel supplied from the motion estimator 200, and if they are equal, outputs a logical "high" signal to it. Otherwise, it sends a logical "low" signal to AND gate 520.

The AND gate 520 has comparators 510, 5
15 in response to the output signal from the comparators 510, 515.
Generate a logic "high" signal as the control signal CTL to the terminal S1 of the switch 530, and a logic "high" signal otherwise.

According to the invention, the device of the invention further comprises a vector selection block 600, which of the first and second main vectors has a smaller error function (for example, an average of 2). Select the dominant vector that yields the motion-compensated handling block with power error). This vector selection block 600 is the predictor 6
10, 615 and mean square error (MSE) detector 62
It consists of 0 and 625.

In particular, as shown in FIGS. 1 and 2,
The first and second main vectors are the majority detector 40.
0 to the predictors 610 and 615, respectively. Each predictor 610, 615, in response to each major vector, retrieves the prediction data of the serving block (ie, the pixel data of the previous frame corresponding to the first and second major vectors) from the frame memory on line 110. , And sends the data to the MSE detectors 620 and 625, respectively. This MS
The E detectors 620, 625 calculate the mean squared error between their two inputs (ie, the serving block data of the current frame and its corresponding prediction data provided through line 105) and the calculated value ( That is, the mean square errors E1 and E2) are sent to the comparator 630, respectively.

Comparator 630 compares its two input values (ie, E1 and E2) and has a two input terminal coupled to the select signal to output terminals 400A, 400C of majority detector 400. (MU
X) to 635. If E1 is greater than or equal to E2, the comparator 630 sends a logic “high” select signal to the MUX.
(Multiplexer) 635 to provide the first main vector on line 605 to terminal S2 of switch 530, otherwise to the second main by sending a logic "low" select signal to MUX 635. The vector is supplied to the terminal S2.

The switch 530 responds to the control signal CTL to select one of the two input signals input to its two terminals S1 and S2. In other words, the switch 530 selects the input of the terminal S1 (that is, the motion vector of the center pixel) as the motion vector of the handling block when the control signal CTL is a logic “high” signal,
If the control signal CTL is a logic "low" signal, then the input at terminal S2 (i.e., the main vector that produces the smaller mean square error) is selected as the motion vector for the handling block.

While we have described specific embodiments of the present invention, those skilled in the art will be able to make various modifications without departing from the scope of the invention.

[0021]

Therefore, according to the present invention, when the motion vector representing the displacement between the current frame and the previous frame of the video signal is specified, the motion vector detected through the block matching is filtered through the median filter. , The optimum motion vector is selected from the filtered motion vector and the center motion vector.
Therefore, the motion vector can be accurately specified.

[Brief description of drawings]

FIG. 1 is a block diagram showing a portion of an apparatus of the present invention for identifying a motion vector between a current frame and a previous frame of a video signal.

FIG. 2 is a block diagram showing part of an apparatus of the present invention for identifying a motion vector between a current frame and a previous frame of a video signal.

FIG. 3 illustrates a filtering window according to the present invention for median filtering of pixels.

[Explanation of symbols]

 100 frame memory 105, 110 line 200 motion estimator 300 median filter 400 majority detector 400A to 400C output terminals of majority detector 510, 515 comparator 520 AND gate 530 switch 600 vector selection block 610, 615 predictor 620, 625 2 Multiply Mean Error Detector (MSE) 630 Comparator 635 Multiplexer (MUX)

Claims (3)

[Claims] 1. Used in a motion compensated video encoder,
A motion vector that represents a displacement between a current frame and a previous frame of a video signal is specified. The current frame is divided into a plurality of handling blocks having P × Q pixels, and the P × Q pixels are each divided. A motion vector detector (P and Q are integers greater than 2) comprising a center pixel representing the pixel located at the center of the handling block, which provides a motion vector to each pixel of the current frame, A motion vector providing unit, the motion vector comprising a center motion vector for each center pixel, and a median filtering unit for median-filtering the motion vector to generate a filtered motion vector for a pixel included in the current frame. And the filtered motion vectors corresponding to the pixels included in each handling block. By counting the number of the filtered motion vectors having the same value between the cutouts, the counted first primary vector of each of the handling blocks and the second largest secondary vector of the number of processed blocks are counted. Selecting means for selecting vectors, calculating an error function for each of the handling blocks based on the first and second main vectors, and outputting a vector that produces a minimum error function in the first and second main vectors Selecting means for selecting as a motion vector; control signal generating means for generating a control signal based on the counted number of the center motion vector, the first main vector and the first main vector for each handling block, In response to the control signal, a motion for each handling block is generated from the output motion vector and the center motion vector. The motion vector detection apparatus; and a switching means for selecting a vector. 2. The median filtering means,
The motion vector detection device according to claim 1, comprising a median filter having a cross-shaped filtering window for averaging the motion vectors. 3. Used in a motion compensated video encoder,
A motion vector that represents a displacement between a current frame and a previous frame of a video signal is specified. The current frame is divided into a plurality of handling blocks having P × Q pixels, and the P × Q pixels are each divided. A motion vector detector (P and Q are integers greater than 2) comprising a center pixel representing the pixel located at the center of the handling block, which provides a motion vector to each pixel of the current frame, A motion vector providing unit, the motion vector comprising a center motion vector for each center pixel, and a median filtering unit for median-filtering the motion vector to generate a filtered motion vector for a pixel included in the current frame. And the filtered motion vectors corresponding to the pixels included in each handling block. By counting the number of the filtered motion vectors having the same value between the cutouts, the counted first primary vector of each of the handling blocks and the second largest secondary vector of the number of processed blocks are counted. Sorting means for selecting vectors, calculating an error function for each of the handling blocks based on the first and second main vectors, and outputting a vector that produces a minimum error function in the first and second main vectors Selecting means for selecting as a motion vector, and in response to the center motion vector and the first main vector, generates a first control signal when the center motion vector and the first main vector are equal, and otherwise ,
Comparing the number of the first main vectors with a predetermined number and generating a second control signal, and if the number of the first main vectors is greater than or equal to the predetermined number, Second control means for generating a third control signal, otherwise generating a fourth control signal; and, in response to the first, second, third and fourth control signals, the first and third control signals. Is applied, the third selection means generates a first selection signal, and otherwise generates a second selection signal, the center motion vector responsive to the first selection signal, and the second selection signal. And a switching means for selecting a motion vector for each handling block among the output motion vectors in response to the motion vector detecting device.