KR100255634B1 - Motion detecting apparatus - Google Patents

Abstract

PURPOSE: A motion detector is provided to minimize the degradation of image quality, and to simplify hardware of a motion estimation of an encoder by selectively performing an 1/2 pel motion estimation. CONSTITUTION: An 1 pel motion estimation block(10) estimates image data as the unit of 1 pel, and generates 1 pel motion vector. A judgement block(40) generates 1/2 pel motion vector value as 0 in case that a motion estimation as the unit of 1/2 pel isn't needed. The judgement block(40) outputs data passed through the 1 pel motion estimation block in case that the motion estimation is needed. An interpolation block(20) inputs the data provided from the judgement block(40). An 1/2 pel motion estimation block(30) generates 1/2 pel motion vector.

Description

Motion detector

1 is a block diagram of an encoder of a video device in general MPEG1.

2 is a block diagram showing a conventional configuration of the motion detection device of FIG.

3 is a view for explaining a half-pel motion evaluation.

4 is a block diagram of a motion detection apparatus according to the present invention.

5 is a diagram for explaining a concept of interframe motion evaluation.

6 is a block diagram of one embodiment of a motion detection apparatus according to the present invention.

The present invention relates to a motion detection device, and more particularly to a motion detection device that can simplify the circuit configuration.

Digital image compression technology for the transmission and recording of digital moving images has attracted much attention in the world in recent years, and one of the results is unifying global standards through the Moving Picture Expert Group (MPEG) activity. Already, MPEG-1, which is aimed at VHS quality, is in practical use, and MPEG-2 of high quality is being put into practical use.

In general, video apparatuses employing the algorithm of MPEG1 perform motion compensation in a decoder and motion evaluation and compensation in an encoder as shown in FIG. Since motion evaluation as well as motion compensation is very complicated in its hardware implementation, the performance and price of the device largely depend on how the hardware of the motion evaluation is implemented.

In the conventional motion estimation, a 1-pel motion vector is obtained from a 1pel motion evaluation block 10 from the input image data as shown in FIG. The interpolated block 20 is interpolated to obtain surrounding half-pel pixels at the position, and the half-pel motion vector is obtained from the half-pel motion evaluation block 30. By doing this, possible motion vectors in the search area have a precision of up to one decimal place. That is, as shown in FIG. 3, first, after evaluating the motion of the integer pel unit, the half pel value around the pel corresponding to the motion vector found by the interpolation filter is extracted, and then the motion vector of one digit after the decimal point is second-order. You will find it.

By doing so, it is possible to reduce the prediction error through motion compensation and improve the encoding efficiency by obtaining the motion vector up to one decimal place, but the motion evaluation itself requires a lot of hardware and furthermore, the half-pel motion. There was a problem that the implementation was not easy to find the motion by the evaluation block 30.

SUMMARY OF THE INVENTION An object of the present invention is to provide a motion detection apparatus capable of simplifying hardware of motion estimation of an encoder while minimizing image degradation by selectively performing 1 / 2-pel motion evaluation to solve such problems of the prior art. .

In order to achieve the above object, the apparatus of the present invention is a motion detection device for image data, comprising: a 1-pel motion evaluation block for generating a 1-pel motion vector by estimating the motion of the image data in 1-pel units; A determination block for determining whether to perform a motion evaluation of a half pel unit according to the motion evaluation result of the one pel unit; An interpolation block for extracting 1/2 pel values around the position of the 1 pel motion; A half pel motion evaluation block for generating a half pel motion vector by performing a half pel motion evaluation from the extracted half pel values; And output means for selectively outputting the output of the 1/2 pel motion evaluation block according to the determination result of the determination block.

If the minimum value of the displaced block difference value obtained from the 1-pel motion evaluation block is zero, the judging means does not perform 1 / 2-pel motion evaluation, and otherwise.

In addition, the determination means may not perform the half-pel motion evaluation when the difference between the maximum value and the minimum value of the displaced block difference value obtained in the one-pel motion evaluation block is less than the average value, and others may be performed.

In addition, the determining means may not perform the half-pel motion evaluation if the degree of change of the displaced block difference value obtained in the one-pel motion evaluation block is less than or equal to the minimum value, and others may be performed.

Accordingly, the present invention can reduce the complexity of the hardware while minimizing coding efficiency and quality degradation after reconstruction by adjusting the degree of motion vector to 1 pel or 1/2 pel.

Hereinafter, with reference to the preferred embodiment shown in the accompanying drawings will be described in detail the present invention.

4 is a block diagram of a motion detection apparatus according to the present invention. The motion detection apparatus includes a 1-pel motion evaluation block 10 that generates a 1-pel motion vector by evaluating the motion of the image data in 1-pel units, and a motion of 1 / 2-pel unit is evaluated according to the motion evaluation result of 1-pel unit. If it is not necessary, the half-pel motion vector is generated as zero, and if there is, the determination block 40 for outputting the data passing through the one-pel motion evaluation block and the data supplied through the determination block 40 are inputted. The interpolation block 20 extracts the half-pel values around the one-pel movement at the position where the one-pel movement is evaluated, and the half-pel movement is performed by performing the movement evaluation in units of the half-pel unit from the extracted half-pel values. It includes a half-pel motion evaluation block 30 for generating a vector.

That is, the present invention first evaluates the 1-pel movement and determines whether or not to perform the 1 / 2-pel movement evaluation according to a given condition. If it is determined that it should be performed, the half-pel motion vector is obtained by performing the interpolation and 1 / 2-pel motion evaluation according to the conventional method. Just to complete the movement evaluation.

Referring to FIG. 5, the inter-frame motion estimation has a low coding efficiency when only the difference between the previous frame and the current frame is coded. Therefore, the difference is estimated by predicting the motion to reduce data of the difference and compensating the motion. To reduce the data effectively. That is, as shown in FIG. 5, to find the position of the candidate block having the smallest Displaced Block Difference (DBD) among the many candidate blocks in the constant search window of the previous frame with respect to the reference block in the current frame. Motion vector. In this case, the DBD value can be obtained by various methods, but usually, a MAE (Mean Absolute Error) is obtained by the following equation.

The above equation represents the sum of the absolute difference between the data Y of the previous frame and the data X of the current frame having the block size of M and N, respectively, and the method for finding all the candidate blocks in the search window is called Full Search. Several methods have been proposed, such as logarithmic search, to obtain only a few and reduce the amount of computation.

In the present invention, first, it is possible to determine whether or not to obtain a motion vector of about 1/2 pel using the DBD value obtained at about 1 pel.

In the first method, if the DBD value is determined as zero, the DBD value is zero, which means that the motion evaluation is accurate. Therefore, the optimal state has already been achieved. Therefore, it is not necessary to obtain the motion evaluation of about 1/2 pel. Therefore, if the minimum value of the DBD is zero, the half-pel motion vector is generated as zero and half-motion evaluation is not performed.

Second, the DBD minimum value is not zero, but when the difference between the DBD minimum value and the DBD maximum value does not exceed the DBD average value, it means that there are many similar blocks in the search window because the distribution of the DBD value is distributed in a narrow area. Based on the determination of the distribution of the DBD values, the 1-pel motion evaluation generates zero half-pel motion vectors without evaluating the half-pel motion in consideration of image quality effects and hardware complexity. If the DBD average value or more, 1/2 pel motion evaluation is performed as in the prior art.

The third method calculates the degree of change of the DBD value and if the degree of change is less than the minimum DBD value, the half-pel motion vector is generated zero without performing the half-pel motion evaluation. Do this.

6 shows a block diagram of a preferred embodiment of the motion detection apparatus according to the present invention. An embodiment of the motion detection apparatus includes a 1-pel motion evaluation block 10 for generating a 1-pel motion vector by evaluating motion of the image data in 1-pel units, and a 1 / 2-pel unit according to the motion evaluation result of 1-pel unit. A determination block 50 for determining whether to perform motion evaluation, an interpolation block 20 for extracting 1/2 pel values around the position where the 1 pel motion is evaluated, and 1 from the extracted 1/2 pel values A half-pel motion evaluation block 30 that generates a half-pel motion vector by performing a motion evaluation of a half-pel unit; And output means 60 for selectively outputting the output of the 1/2 pel motion evaluation block according to the determination result of the determination block.

Accordingly, the present invention can reduce the complexity of the hardware while minimizing coding efficiency and quality degradation after reconstruction by adjusting the degree of motion vector to 1 pel or 1/2 pel.

Claims (5)

An apparatus for detecting motion of video data, comprising: a 1-pel motion evaluation block for generating a 1-pel motion vector by evaluating the motion of the image data in units of 1-pel; The accuracy or distribution of the motion estimation is determined by using a predetermined method for the displaced block difference value of candidate blocks in the search window of the previous frame with respect to the reference block in the current frame in the unit of the 1-pel motion evaluation block. A judging block for judging to perform the motion evaluation of a half pel unit only when the predetermined reference value or more is determined after the judgment; An interpolation block for extracting 1/2 pel values around the position of the 1 pel motion; A half pel motion evaluation block for generating a half pel motion vector by performing a half pel motion evaluation from the extracted half pel values; And output means for selectively outputting the output of the 1/2 pel motion evaluation block according to the determination result of the determination block. The motion detection apparatus of claim 1, wherein the determination block does not perform a half-pel motion evaluation when the minimum value of the displaced block difference value obtained in the one-pel motion evaluation block is zero, and otherwise. The method of claim 1, wherein the determination block does not perform a half-pel motion evaluation when the difference between the maximum value and the minimum value of the displaced block difference value obtained from the one-pel motion evaluation block is less than the average value, and the rest is performed. Motion detection device. The motion detection method of claim 1, wherein the determination block does not perform a half-pel motion evaluation if the variation of the displaced block difference value obtained from the one-pel motion evaluation block is less than or equal to a minimum value, and performs the others. Device. An apparatus for detecting motion of video data, comprising: a 1-pel motion evaluation block for generating a 1-pel motion vector by evaluating the motion of the image data in units of 1-pel; The accuracy or distribution of the motion estimation is determined by using a predetermined method for the displaced block difference value of candidate blocks in the search window of the previous frame with respect to the reference block in the current frame in the unit of the 1-pel motion evaluation block. A judgment block for generating a half pel motion vector value as zero after the determination and outputting data having passed the one pel motion evaluation block if it is less than a predetermined reference value; An interpolation block for inputting data supplied through the determination block to extract 1/2 pel values around the 1 pel movement from the position where the 1 pel movement is evaluated; A half pel motion evaluation block for generating a half pel motion vector by performing a half pel motion evaluation from the extracted half pel values; And output means for selectively outputting the output of the 1/2 pel motion evaluation block according to the determination result of the determination block.