KR100672959B1 - Apparatus and method for fast encoding for video - Google Patents

Abstract

A fast video encoding apparatus and a method thereof are provided to omit a process of calculating an unnecessary motion vector by determining a compression mode before a motion vector is calculated to reduce the quantity of calculations and time required for calculations. A fast video encoding apparatus includes a sum calculator(300), a controller(310), a motion vector calculator(320), and an output unit(330). The sum calculator calculates a first sum value corresponding to the sum of pixel values of a predetermined to-be-compressed object unit region of object video information to be compressed and a second sum value corresponding to the sum of pixel values of a predetermined compression reference unit region of reference video information. The controller outputs a motion vector calculation signal when the difference between the first sum value and the second sum value is greater than a predetermined threshold value. The motion vector calculator calculates a motion vector with respect to the to-be-compressed object unit region when receiving the motion vector calculation signal from the controller. The output unit outputs the motion vector received from the motion vector calculator.

Description

Apparatus and method for fast encoding for video}

1 is a diagram schematically illustrating a method of obtaining a motion vector through a conventional motion search technique used in the related art.

2 is a flowchart illustrating a process of compressing image information through a motion search method of a macroblock used in the related art.

 3 is a block diagram showing a configuration of a preferred embodiment of a high speed video encoding apparatus according to the present invention;

4 is a flowchart illustrating a process of compressing image information by a high speed video encoding method according to the present invention.

5 is a view showing an experimental result comparing the operation processing time for the number of frames by the compression method of the prior art and the compression method according to the present invention.

6 is a diagram illustrating a conventional compression method and a compression method according to the present invention in contrast to a noise signal ratio with respect to the number of frames of an original image.

The present invention relates to an encoding apparatus and method in a digital video recorder. More particularly, by first determining a mode for a macroblock, which is a compression target unit area, and then searching for a motion vector, it is possible to reduce unnecessary search and reduce complexity of computation. A high speed video encoding apparatus and method.

As an example of a video recorder, a DVR (Digital Video Recorder) system converts analog video signals of multiple channels input to a surveillance camera into digital, and stores, plays back and retrieves them using video compression technology. It is a system that can monitor and control video remotely.

The DVR system can replace the AVR (Analog Video Recorder), which has been used at a high speed, because it can store video signals semi-permanently in a storage device, can easily manage the storage device, and can receive streaming services in real time. .

Images of each channel obtained through the video camera are encoded using image compression technology and recorded in a storage device or transmitted to the user through the channel.

In such a system, analog video signals are converted to digital signals and video compression techniques are used. The video compression standards (MPEG, H.26x) for these compression techniques use high compression ratios by eliminating temporal redundancy, spatial redundancy, and statistical redundancy. We adopt method to get.

A compression method using spatial and statistical redundancy is a method of dividing a single image into macroblock units and compressing them through DCT (Discrete Consine Transform), quantization, and variable length coding. In general, the compression efficiency is lower than the method of eliminating temporal redundancy, but it has the advantage of maintaining good image quality even after the decoding process.

In contrast, temporal redundancy has the most data redundancy in successive video sequences. The way to remove this is mainly through motion vector estimation.

Since the motion search module occupies 67% or more of the total complexity of the encoder, the optimization of the motion search module is essential to improve the complexity of the encoder. Especially, in the case of multi-channel coding such as a DVR, the complexity increases considerably. Increased occupancy causes problems with system stability.

Hereinafter, a general motion search technique conventionally used will be described.

FIG. 1 is a diagram schematically illustrating a method of obtaining a motion vector through a conventional motion search technique used in the related art.

Referring to FIG. 1, the basic concept of calculating a motion vector used in a method of compressing image information by removing temporal redundancy is that one frame does not transmit an image per frame for a still image of about 30 frames per second. The difference between the still image of the still image and the previous still image and the image information between the current still image and the data is transmitted, and the method of compensating and restoring in the process of decoding again.

In order to calculate a motion vector representing the difference between the current still image and the still image of the previous frame, a target macroblock region, which is a region of compression, is set among the current still images, and the block most similar to the macroblock is referred to. The search is performed in units of blocks equal to the size of the target macroblock in the search region of the previous image.

As a macroblock corresponding to the target macroblock in the reference image, a macroblock matched with the target macroblock and compared is defined as a reference macroblock.

When the most similar reference macroblock in the search area is found, the moving coordinates between the blocks are converted into data by compressing the current target macroblock and the reference macro in the search area.

The search for the search region of the reference image is divided into macroblock units, and a block matching method for the reference image, that is, the macroblock of the current frame to be encoded, that is, the reference macroblock most similar to the target macroblock is found. Use the method to find it.

The equation for the SAD (Sum of Absoluted Difference) function, which is the most common function for finding the optimal motion vector used for block matching, that is, the most similar reference macroblock, is shown in Equation 1 below.

In the above formula, M and N are the width and length of the macro block, and C (i, j) is the pixel value of the i, j coordinate of the current macroblock to be compressed, and R (i + m, j + n) denotes pixel values of the i + m, j + n coordinates of the reference macroblock corresponding to the macroblock to be compressed in the search region of the reference image, and m, n represent candidate motions as coordinate values of the search region. It becomes a vector. Among the m and n, the m and n values such that the SAD (m, n) function has the smallest value become the optimal motion vector. This is summarized in the following [Equation 2].

In the above formula, S denotes a search area, and in general, ± 16 magnitudes from the center in the case of full search. As a result, (Ix, Iy), which is the value of (m, n) to have the smallest SAD ((m, n), becomes a motion vector by the above equation (2).

A flowchart of a conventional method of compressing an image using the motion vector after obtaining the motion vector through Equations 1 and 2 is shown in FIG. 2.

2 is a flowchart illustrating a process of compressing image information through a motion search method of a macroblock used in the related art.

Referring to FIG. 2, a motion vector is calculated in advance by compressing the image information by the method described in FIG. 1 (S200).

When it is determined that the calculated motion vector is smaller than the predetermined reference value (S210), that is, when it is determined that there is almost no difference between the reference macroblock and the target macroblock, the compression method by the motion vector is skipped (S220).

Since the skipped target macroblock is compressed using only information that is the same as the reference macroblock, the skipped macroblock can be compressed with a much smaller amount of data.

If it is determined that the difference between the reference macroblock and the target macroblock is larger than the reference value, whichever is more efficiently, the difference between the macroblock of the original image and the macroblock of the original image and the motion compensated macroblock is used. It is determined whether the data can be compressed (S230).

S230 and below S240 and S250 are well-known techniques well known to those skilled in the art, and detailed descriptions thereof are omitted because they are not intended to be protected by the present invention.

In the prior art, the calculation of the motion vector is performed first and the skip is determined accordingly. Therefore, a large amount of computation is required for the calculation of the motion vector. Quite high complexity is required.

In addition, since the operation time and overload are particularly important in a device and a method for receiving signals for multiple channels such as a DVR, there are many problems when the conventional method is used as it is.

The technical problem to be achieved by the present invention is to avoid the determination of skipping using a motion vector to solve the problems of the prior art as described above, and to become a compression target by using the characteristics of the image that the motion is not large. By comparing the absolute value of the sum of the pixel values of the target macroblock and the sum of the pixel values of the reference macroblock, it is possible to determine whether to skip or reduce the amount of calculation. It is to provide an apparatus and method that can reduce the amount of computation through partial operation.

Another object of the present invention is to compare the absolute value of the sum of the pixel values of the target macroblock to be compressed and the sum of the pixel values of the reference macroblock to determine whether or not to reduce the amount of computation and to significantly reduce the amount of motion. In computing a vector, a computer-readable recording medium having stored thereon a program for executing a method in a computer can be used to avoid computing a whole area and to reduce the amount of computation through partial operations.

According to an aspect of the present invention, there is provided a high speed video encoding apparatus according to an embodiment of the present invention, comprising: a first sum value, which is a sum of pixel values of a predetermined compression target unit region, of a target image information to be compressed; A sum value calculation unit configured to calculate a second sum value that is a sum of pixel values of a compression reference unit area, which is a predetermined area of reference image information, which is a reference of compression of the corresponding target image information; A control unit which outputs a motion vector calculation signal when the difference between the first sum value and the second sum value input from the sum value calculation unit is greater than a predetermined threshold value; A motion vector calculation unit for calculating a motion vector for the compression target unit region by a predetermined method when a motion vector calculation signal is input from the controller; And an output unit for outputting a motion vector received from the motion vector calculation unit.

According to another aspect of the present invention, there is provided a fast video encoding method according to an embodiment of the present invention, wherein a pixel value of a predetermined compression target unit region among the target image information to be compressed and the target image information corresponding to the compression target unit region are provided. A motion vector calculation determining step of determining whether or not to calculate a motion vector by a predetermined method using a difference of pixel values of a compression reference unit area, which is a predetermined area of reference image information, which is a reference of compression for the? And a motion vector calculation step of calculating a motion vector for the compression target unit region by a predetermined method when it is determined that the motion vector is calculated in the motion vector calculation determination step.

Thus, before calculating the motion vector, the block mode is first determined by using the pixel value of the compression reference unit region so that the motion search for the skipped macroblock can be omitted, thereby reducing the computation time and the motion search. In calculating the motion vector, it is possible to implement a high speed video encoding that avoids the operation of the global search area and minimizes the effect of the image quality and reduces the computation time even through only a partial calculation.

Hereinafter, exemplary embodiments of a fast video encoding apparatus and method according to the present invention will be described in detail with reference to the accompanying drawings.

 3 is a block diagram showing a configuration of a preferred embodiment of a high speed video encoding apparatus according to the present invention.

Referring to FIG. 3, the high speed video encoding apparatus according to the present invention includes a sum value calculating unit 300, a control unit 310, a motion vector calculating unit 320, and an output unit 330.

The sum value calculation unit 300 is a reference value for compression of the target image information corresponding to the first sum value and the compression target unit region, which is the sum of the pixel values of the predetermined compression target unit region of the target image information to be compressed. A second sum value, which is the sum of pixel values of the compression reference unit area, which is a predetermined area of the reference image information, is calculated.

The sum value calculating unit 300 performs arithmetic processing in units of macroblocks, which are predetermined compression target unit areas, of the target image information to be compressed.

A macroblock is a unit that is determined by dividing a predetermined area in an image to be compressed, and means a unit area that is a standard of operation in compressing image information. In general, a macroblock includes 256 (16 × 16) pixels. .

The sum value calculator 300 sums pixel values of respective pixels forming a target macroblock formed by dividing the target image information to be compressed. The pixel value of each pixel is mainly composed of 8 bits. A first summation value, which is the sum of all pixel values of 256 pixels forming a normal macroblock size of 16 × 16, is calculated.

The second sum value is calculated by calculating the pixel value of the reference macro block corresponding to the target macro block from the first sum value in the corresponding reference image information.

Table 1 will be described in detail with respect to the method for calculating the first and second total value.

Reference image MB _r0 MB _r1 MB _r2 MB _r3 MB _r4 MB _r5 MB _r6 MB _r7 MB _r8 MB _r9 … … … … … … … … … … … …

Current video

MB _c0 MB _c1 MB _c2 MB _c3 MB _c4 MB _c5 MB _c6 MB _c7 MB _c8 MB _c9 … … … … … … … … … … … …

In Table 1, MB _r0 (MacroBlock reference0) means the 0th macroblock of the reference image, and MB _c0 (MacroBlock current 0) means the 0th macroblock of the image currently being compressed. It consists of 16 pixels.

As shown in Table 1, the sum values of the conventional 256 pixels constituting each macroblock of the reference image and each macroblock of the current image are calculated. The difference between the first sum value, which is the calculated sum of the macroblocks of the current image, and the second sum value, the calculated sum of the macroblocks of the reference video, is calculated.

The controller 310 outputs a motion vector calculation signal to control to calculate a motion vector only when the difference between the first sum value and the second sum value input from the sum sum calculation unit 300 is greater than a predetermined threshold value. do.

The threshold is a value obtained empirically through experiments, and the value is determined based on the place where the device to which the idea of the present invention is implemented, the content of image information, the processing power of the central processing unit, the number of image input channels, etc. as variables. Becomes Experimental results in the detailed description of the present invention was tested by setting the threshold value to 5. That is, when the sum of pixel values of the pixels constituting the target macroblock is less than 5 or greater than -5 than the sum of the pixel values of the pixels constituting the reference macroblock, the target macroblock to be compressed is coincident with the reference macroblock. To judge.

In the prior art, the motion vector is calculated first, and the skip vector is used to determine whether to skip the motion vector. However, the present invention requires the calculation of the motion vector in advance. By calculating the motion vector by the method, a large amount of computation can be reduced.

That is, the method employed in the present invention, unlike the prior art, the difference between the sum of the pixel values of the pixels constituting the target macroblock and the reference macroblock is less than the predetermined threshold value is almost the current target macroblock and the reference macroblock. It will be used as a standard that can be judged to match.

In the case of video information for DVR, the change of motion is not large and the necessity of guaranteeing the fineness of image quality is rather low. Compression can be compressed.

If the difference between the first sum value and the second sum value is smaller than a predetermined threshold value, it is determined that the target macroblock matches the reference macroblock and is compressed in the skip mode. In this case, the compression of the skip mode is compressed to only information indicating that the current target macroblock matches the reference macroblock. Since only 1 bit of information is sufficient, this information compresses the target macroblock consisting of 256 pixels into only 1 bit of information.

After comparing MB _r0 and MB _c0 by the above method, the sum of the pixels of MB _r1 and MBc1 is calculated and the above operation is performed. In this order, it is determined whether all the compression target macroblocks are skipped. do.

The controller 310 determines whether to obtain motion vectors of all macroblocks of the target image or compress them in the skip mode in the same manner as described above.

Since the motion vector is calculated only when the difference between the first sum value and the second sum value is larger than the predetermined threshold value, the controller 310 determines that the difference between the first sum value and the second sum value is greater than the predetermined threshold value. In this case, a motion vector calculation signal is output to control to calculate a motion vector.

When the motion vector calculation signal is input from the controller 310, the motion vector calculation unit 320 calculates a motion vector for the target macroblock by a predetermined method.

Since a general method for calculating a motion vector has been described in detail above with reference to FIGS. 1 and 2, a method for calculating a motion vector according to the present invention will be described below.

A method of calculating a motion vector for a target macroblock first sets an area within a predetermined range as a search region around the region of the reference macroblock corresponding to the target macroblock in the reference image.

By comparing pixel values between the reference macroblock corresponding to the area having the same size as the target macroblock and the target macroblock in the search area, the reference macroblock in the most similar search area is determined, and the reference macroblock in the determined search area The difference between the positions of the target macroblocks is set to two-dimensional coordinates, and the set two-dimensional coordinates become motion vectors.

In a conventional method, a motion vector is calculated by comparing all pixel values between a target macroblock and a macroblock that may be all candidate vectors in a search region, but in the present invention, all the corresponding pixel values are not calculated to reduce the amount of computation. A method of calculating only pixel values of columns and rows having a predetermined interval is employed.

An example of a 16 × 16 macroblock commonly used will be described.

If the predetermined interval is set to odd columns and odd rows, in comparing pixel values of the target macroblock and the reference macroblock corresponding to the target macroblock in the search region, the (1,1) of the target macroblock and the reference macroblock are compared. With pixel values of (1,3), (1,5), (1,7) .....

(3,1), (3,3), (3,5), ....... (5,1), (5,3), (5,5) ..... The same operation is performed again by moving to the next reference macroblock.

Through this operation, the reference macroblock corresponding to the target macroblock closest to the target macroblock is found. If the above-mentioned operations using only odd columns and odd rows are expressed by Equation 3, Equation 3 is given.

Since Equation 3 has been described through Equation 1, a detailed description thereof will be omitted.

When the calculation is performed by the method employed in the present invention, the amount of calculation is much smaller in calculating the motion vector than in the case of calculating the entire area of the search area.

The output unit 330 outputs the motion vector calculated by the motion vector calculation unit 320 as described above.

Hereinafter, a high speed video encoding method according to the present invention will be described in detail with reference to FIG. 4.

4 is a flowchart illustrating a process of compressing image information by a high speed video encoding method according to the present invention.

In order to compress the image information by the fast video encoding method according to the present invention, the sum value calculating unit 300 first calculates a first sum value that is the sum of pixel values of the target macroblock in the compression target unit area (S400).

In addition, the sum value calculation unit 300 calculates a second sum value that is the sum of pixel values of the reference macro block, which is the compression reference unit area in the reference image corresponding to the target macro block (S410).

In this way, the controller 310 receives the first sum value and the second sum value calculated by the sum value calculating part 300 and determines whether the difference between the two values is greater than a predetermined threshold value (S420).

The threshold is a value obtained empirically through experiments, and the value is determined by using a variable such as the place where the device for implementing the idea of the present invention is installed, the content of the image information, the size of the central processing unit, the number of channels, and the like. Experimental results in the detailed description of the present invention was tested by setting the threshold value to 5.

That is, when the sum of pixel values of the pixels constituting the target macroblock is less than 5 or greater than -5 than the sum of the pixel values of the pixels constituting the reference macroblock, the target macroblock to be compressed is coincident with the reference macroblock. To judge.

If the difference between the first sum and the second sum is greater than the threshold, the motion vector calculator 320 calculates a motion vector (S430).

In addition, when the difference between the first sum and the second sum is smaller than the threshold, the control unit 310 does not compress the motion vector, but compresses only the information that matches the reference macroblock (S440).

Hereinafter, the results of the comparative experiment on the performance of the technical concept of the present invention and the prior art will be described with reference to FIGS. 5 and 6.

First, Table 2 shows the experimental results of the performance of the method according to the conventional video compression standard MPEG-4 and the method according to the present invention with respect to the time spent coding the test image having 300 frames.

QCIF (176 × 144) test image was used in the experiment for explaining the present invention in the detailed description of the present invention.

consumed time Time / frame Frames / hour Noise ratio MPEG-4 2273.40 ms 7.578 ms 131.96 41.83 dB Invention 1206.60 ms 4.022ms 248.63 40.95 dB

As can be seen from Table 2, when the compression method according to the present invention reduces the time required to calculate 300 frames in half compared with the conventional method, the noise ratio is 0,88dB when compared with the conventional method. The difference is not so old that the performance is almost the same in image quality.

Considering the fact that the motion, which is a characteristic of the DVR video image, is not large, even if the video compression method according to the present invention is compared with the original image and the image quality, the amount of computation and time can be further reduced without almost degrading performance.

5 is a view showing an experimental result comparing the operation processing time for the number of frames by the conventional compression method and the compression method according to the present invention.

Referring to FIG. 5, in the method of the present invention, the distribution of the processing time according to the calculation for each frame is kept constant without having a large peak value, and the processing time is less than half as compared with the prior art. It can be seen that the computation time is reduced. On the contrary, according to the prior art, a large amount of calculations are suddenly increased according to a frame, and a large amount of peak distributions may be found, which may cause anxiety of the system.

According to the present invention, it is shown in Figure 6 the results of experiments what effect on the image quality compared to the prior art.

6 is a diagram illustrating a conventional compression method and a compression method according to the present invention in contrast to a noise signal ratio with respect to the number of frames of an original image.

Referring to FIG. 6, a frame having a peak distribution is compared with a noise signal by a conventional compression method, but generally has a distribution almost similar to a noise signal by a conventional compression method. This suggests that the method according to the present invention has little effect on image quality as compared with the method adopted in the prior art.

The invention can also be embodied as computer readable code on a computer readable recording medium. Computer-readable recording media include any type of recording device that stores data that can be read by a computer system. Examples of computer-readable recording media include ROM, RAM, CO-ROM, magnetic tape, and floppy. Disks, optical data storage, and the like, and also include those implemented in the form of carrier waves (eg, transmission over the Internet). The computer readable recording medium can also be distributed over network coupled computer systems so that the computer readable code is stored and executed in a distributed fashion.

Although the preferred embodiments of the present invention have been shown and described above, the present invention is not limited to the specific preferred embodiments described above, and the present invention belongs to the present invention without departing from the gist of the present invention as claimed in the claims. Various modifications can be made by those skilled in the art, and such changes are within the scope of the claims.

In the high speed video compression apparatus and method according to the present invention, the calculation process of calculating unnecessary motion vectors can be omitted by determining the compression mode before calculating the motion vectors, thereby reducing the amount of computation and the computation time. In calculating the motion vector, the computation amount can be reduced by avoiding global calculation of the search area and calculating only a part of the motion vector.

In addition, despite significantly reducing the amount of computation and computation time, the quality of the compressed image can also be maintained at almost the same level as in the prior art, so that the share of the central processing unit (CPU, etc.) can be lowered and the system can be stably operated.

Claims (6)

A predetermined area of reference image information that is a reference for compression of the first sum value that is the sum of pixel values of a predetermined compression target unit region of the target image information to be compressed and the target image information corresponding to the compression target unit region; A sum value calculation unit for calculating a second sum value that is a sum of pixel values of the compression reference unit region; A control unit which outputs a motion vector calculation signal when the difference between the first sum value and the second sum value input from the sum value calculation unit is greater than a predetermined threshold value; A motion vector calculation unit for calculating a motion vector for the compression target unit region by a predetermined method when a motion vector calculation signal is input from the controller; And And an output unit for outputting a motion vector received from the motion vector calculating unit. The method of claim 1, Motion vector calculation unit, Pixel values constituting a corresponding area corresponding to the compression target unit area among the predetermined search areas for the compression target unit area in the reference image information have a predetermined interval among pixel values constituting the compression target unit area. And calculating a motion detector using coordinates of the compression target unit area with respect to the corresponding area having the smallest calculated difference among the predetermined search areas by calculating a difference between pixel values of a column and a row. Pixel value of a predetermined compression target unit area among the target image information to be compressed and a pixel of the compression reference unit area which is a predetermined area of reference image information which is a reference of compression of the target image information corresponding to the compression target unit area. A motion vector calculation determining step of determining whether to calculate a motion vector by a predetermined method using a difference of values; And And a motion vector calculation step of calculating a motion vector for the compression target unit region by a predetermined method when it is determined that the motion vector is calculated in the motion vector calculation determination step. The method of claim 3, The predetermined method of the motion vector calculation determination step, A first summation value calculating step of calculating a first summation value which is a sum of pixel values of the compression target unit region; A second sum value calculation step of calculating a second sum value that is a sum of pixel values of the compression reference unit area; And And a threshold compression determining step of determining to calculate a motion vector only when the difference between the first sum value and the second sum value is larger than a predetermined threshold value. The method of claim 3, The motion vector calculation step, Columns and rows positioned at predetermined intervals among pixels constituting a corresponding region corresponding to the compression target unit region among the predetermined search areas for the compression target unit region in the reference image information; And calculating a motion vector using coordinates of the compression target unit region with respect to the corresponding region having the smallest calculated difference among the predetermined search regions by calculating a difference between pixel values of. A computer-readable recording medium having recorded thereon a program for executing the high speed video encoding method according to any one of claims 3 to 5.

Images