KR20070119469A - Apparatus and method for video encoding using signal separation algorithm based on independent component analysis and apparatus and method for video decoding using the same - Google Patents

Abstract

An apparatus and a method for video encoding, using a signal separation algorithm based on independent component analysis are provided to improve an existing video coding scheme having a limitation in video compression efficiency for each macro block. A centering unit regulates data average of a residual image of a macro block delivered from a DCT transformer unit. A copy/conversion unit copies and transforms the residual image of the macro block adjusted in the centering unit to create a residual image having the same dimension as the dimension of the image before an encoding operation. An image separation unit separates a macro block data approximating to original data from the residual image having the macro block generated in the copy/conversion unit through a bland source separation(BSS) algorithm.

Description

Apparatus and method for video encoding using signal separation algorithm based on independent component analysis and apparatus and method for video decoding using the same}

1 is a configuration diagram of an embodiment of a conventional video encoding and decoding apparatus;

2 is a diagram illustrating an embodiment of a video encoding and decoding apparatus using an independent component analysis based signal separation algorithm according to the present invention;

3 is a block diagram of another embodiment of a video encoding and decoding apparatus using an independent component analysis based signal separation algorithm according to the present invention;

4 is a diagram illustrating an example of a bit stream indicating whether to apply blind signal separation in the decoding apparatus of FIG.

FIG. 5 illustrates an embodiment of a process of synthesizing data in units of macro blocks in an encoding apparatus according to the present invention; FIG.

6 is an exemplary diagram illustrating a process of restoring data in units of macro blocks in a decoding apparatus according to the present invention;

7 is a flowchart illustrating a video encoding method using an independent component analysis-based signal separation algorithm according to the present invention;

8 is a flowchart illustrating a video decoding method using an independent component analysis based signal separation algorithm according to the present invention.

The present invention relates to a video encoding apparatus and method using the independent component analysis based signal separation algorithm, and a video decoding apparatus using the same, and more particularly, to a video signal using the independent component analysis based signal separation algorithm A video encoding apparatus and method, and a video decoding apparatus using the independent separation analysis based signal separation algorithm for synthesizing and compressing two or more macro blocks, and to restore the compressed macro block to a signal close to the original video signal; It's about how.

In the conventional video coding scheme, encoding and decoding are performed for each macro block in which a video frame is divided into small areas.

An encoding and decoding apparatus of a conventional video coding scheme is as follows.

1 is a block diagram of a conventional video encoding and decoding apparatus.

As shown in FIG. 1, a conventional video encoding apparatus performs motion estimation and motion compensation to remove interframe redundancy for each macroblock in a current frame, or performs intraframe redundancy for each macroblock. Intra prediction is performed to remove. After that, if the conventional video encoding apparatus is set to provide the optimal state, the conventional video encoding apparatus obtains a macro block or a residual image block with the current block to be compressed. Then, the conventional video encoding apparatus performs a discrete cosine transform (DCT), quantization, and entropy coding to transfer a motion vector of each macro block to the decoding apparatus.

Meanwhile, the conventional decoding apparatus performs entropy decoding, inverse quantization, and inverse cosine transform on the data received from the encoding apparatus. Then, the conventional decoding apparatus reconstructs an image through filtering by adding the data to a prediction block through motion compensation or intra prediction.

In the conventional video coding method, the value of the quantization level can be adjusted to increase the data compression efficiency of the macroblock during encoding. When the value of the quantization level is increased, the compression efficiency is increased but the image quality is lowered. Thus, the conventional video coding scheme has been studied for a video compression scheme that can increase the compression efficiency while considering the image quality.

As a result, in the conventional video coding scheme, video compression efficiency is improved for each macro block as video compression standards such as MPEG-1 / 2/4 and H.261 / 263/264 have been developed.

As the video compression technology is advanced as described above, the conventional video coding scheme is limited in expecting the improvement of the video compression efficiency of each macro block.

The present invention has been proposed to solve the above problems, by combining two or more macro blocks of a video signal using a signal separation algorithm based on independent component analysis, and compressing the compressed macro block to a signal close to the original video signal. It is an object of the present invention to provide a video encoding apparatus using the independent component analysis-based signal separation algorithm, and a method thereof, a video decoding apparatus using the same and a method thereof.

Other objects and advantages of the present invention can be understood by the following description, and will be more clearly understood by the embodiments of the present invention. It will also be appreciated that the objects and advantages of the present invention may be realized by the means and combinations thereof indicated in the claims.

According to an aspect of the present invention, there is provided a motion estimation means and intra prediction means to which a current frame is delivered, motion compensation means to which a reconstructed frame is delivered, discrete cosine transform means, quantization means, reorder means, entropy coding means, A video encoding apparatus comprising an inverse quantization means connected to a quantization means and an inverse discrete cosine transform means, the residual image block of a macro block obtained by using one of the motion compensation means and the intra prediction means in a current frame. Block synthesizing means for synthesizing block by block and delivering the result to the discrete cosine transform means; And adjusting and copying / converting data averages in the macroblocks synthesized by the block synthesizing means and delivered from the inverse discrete cosine transform means, and restoring the data by means of a blind signal separation algorithm and providing the intra prediction means and the motion compensation means. And implicit signal separating means.

In addition, the present invention provides a video encoding method of sequentially performing motion estimation, motion compensation, intra prediction, discrete cosine transform, quantization, reordering, and entropy coding on a current frame, wherein any of motion estimation / compensation and intra prediction on a current frame is performed. A first step of determining whether to apply a blind signal separation algorithm upon decoding according to video compression efficiency when the residual image block of the macro block is obtained by performing one; A second step of synthesizing each block constituting the residual image block of the macro block when the application of the blind signal separation algorithm is determined during decoding; And a third step of generating and including a bit stream capable of determining whether to apply a blind signal separation algorithm to the synthesized macro block when decoding.

The present invention also provides a video decoding apparatus including an entropy decoding means, a reordering means, an inverse quantization means, an inverse discrete cosine transform means, and a motion compensation means and an intra prediction means, to which an encoded macroblock is transmitted, wherein the inverse discrete cosine Centering means for adjusting the data average in the residual image of the macro block transferred from the converting means; Copying / converting means for copying and converting the residual image of the macroblock adjusted by the centering means to generate the residual image of the macroblock in the same dimension as before encoding; And image separation means for separating the residual image of the macro block generated by the copying / converting means into a macro block approximating the original data according to the blind signal separation algorithm.

In addition, the present invention provides an encoding method for performing entropy decoding, reordering, inverse quantization, inverse discrete cosine transformation, and intra prediction or motion compensation on a residual image of an encoded macro block, and performing inverse discrete cosine transformation. A first step of adjusting a data average for each block from the residual image of the block; A second step of generating a residual image of a macro block having the same dimension as before encoding by copying / converting the residual image of the macro block whose data average is adjusted; And a third step of dividing the residual image of the macro block into a macro block approximating the original data according to the blind signal separation algorithm.

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings, whereby those skilled in the art may easily implement the technical idea of the present invention. There will be. In addition, in describing the present invention, when it is determined that the detailed description of the known technology related to the present invention may unnecessarily obscure the gist of the present invention, the detailed description thereof will be omitted. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

Prior to explaining the present invention, independent component analysis will be described.

The Independent Component Analysis (ICA) is a method of blind source separation (BSS), and has been proposed to cope with a so-called "cocktail party problem". In other words, if you want to talk to a specific person or listen to background music at a party where people's voices, background music, noise, etc. coexist, you have to separate only the sound you want from other sounds to hear only the sound you want. . Here, each of the sounds separated as described above becomes a "source signal" that constitutes the sound of the entire party hall, and indicates what weights the source signals are added to form the mixed sound of the current party hall. Becomes a "mixed matrix".

In this case, the blind signal separation is an algorithm that separates the source signals from the already mixed signal without any information about how the source signals are mixed, and the independent component analysis is performed in estimating the source signals. It is an algorithm that separates the source signals on the assumption that they are statistically independent of each other.

와 혼합행렬

가 선형조합으로 신호

를 성립하면, 수학식 1과 같이 나타낸다.Specifically, the independent unknown source signal

And mixed matrix

Signals in linear combination

If is established, it is expressed as Equation 1.

를 벡터성분

, 신호

를 벡터성분

, 혼합행렬

를

로 나타내면, 상기 수학식 1은 수학식 2와 같이 간단히 나타낼 수 있다.signal

Vector elements

, signal

Vector elements

, Mixed matrix

To

In Equation 1, Equation 1 may be simply expressed as Equation 2.

Equation 2 may be expressed as Equation 3.

가 계산되면, 상기

를

에 곱하여

를 추정 할 수 있다.That is, in Equation 3

Once is calculated, said

To

Multiply by

Can be estimated.

As described above, the independent component analysis is an algorithm for separating the mixed signal into the source signal only by the independent properties of the components without additional information on the mixed signals. Therefore, the present invention artificially synthesizes two signals by using an algorithm that separates the mixed signal of the independent component analysis into the source signal and separates the two signals into two signals, thereby increasing the compression efficiency of video coding while considering the image quality. .

Hereinafter, the present invention will be described in detail with reference to the drawings.

2 is a diagram illustrating an embodiment of a video encoding and decoding apparatus using an independent component analysis based signal separation algorithm according to the present invention.

As shown in FIG. 2, the video encoding apparatus according to the present invention performs motion estimation and motion compensation to remove inter-frame redundancy of the current frame 201 to be encoded, or performs intra prediction to remove intra-frame redundancy. Thereafter, the encoding apparatus obtains a residual block of macroblocks or macroblocks to be currently synthesized by setting an optimal mode among the two modes. Thereafter, the encoding apparatus synthesizes the two residual image blocks (202), and performs the discrete cosine transform, quantization and entropy coding on the synthesized signal and transmits them to the decoding apparatus.

On the other hand, the video decoding apparatus according to the present invention performs entropy decoding, inverse quantization, and inverse discrete cosine transform on the data transmitted from the encoding apparatus, and then remains in close proximity to the power signal for synthesis 202 through the blind signal separation 203. Restore the video signal. Thereafter, the decoding apparatus reconstructs an image by adding the processed data to the prediction block through motion compensation or intra prediction and stores the image in the reconstructed image buffer 204 through filtering.

3 is a diagram illustrating another embodiment of a video encoding and decoding apparatus using an independent component analysis based signal separation algorithm according to the present invention.

As shown in FIG. 3, the video encoding apparatus according to the present invention synthesizes two residual image blocks in case of using blind signal separation to perform discrete cosine transform and quantization, and does not use blind signal separation. In the case of performing discrete cosine transform and quantization, a method is selected according to whether video compression efficiency is effective (301).

Meanwhile, the video decoding apparatus according to the present invention undergoes inverse quantization and inverse discrete cosine transform when video is compressed using implicit signal separation, and reconstructs an image by using implicit signal separation. When video is compressed without using signal separation, the image is reconstructed through inverse quantization and inverse discrete cosine transform (S302).

FIG. 4 is an exemplary diagram illustrating a bit stream indicating whether to apply blind signal separation in the decoding apparatus of FIG. 3.

As shown in FIG. 4, the bit stream is generated after encoding the video to which the blind signal separation is applied in the encoding apparatus.

A "GOP_ICA_Flag bit" 401 indicates whether to apply blind signal separation in units of groups of pictures (GOP) in the bit stream. That is, the "GOP_ICA_Flag bit" 401 indicates whether the encoding apparatus has synthesized data in GOP units. Through this, the decoding apparatus checks the "GOP_ICA_Flag bit" 401 and determines whether to apply blind signal separation after performing inverse quantization and inverse discrete cosine transform.

Thereafter, the decoding apparatus checks the "Macroblock_ICA_Flag bit" 402.

"Macroblock_ICA_Flag bit" 402 indicates whether to apply blind signal separation in units of macro blocks. That is, the "Macroblock_ICA_Flag bit" 402 indicates whether the encoding apparatus has performed data synthesis in units of macro blocks. Through this, the decoding apparatus checks the "Macroblock_ICA_Flag bit" 402 to determine whether to apply blind signal separation after performing quantization and inverse discrete cosine transform on a macroblock basis. As an example, the decoding apparatus checks the "Macroblock_ICA_Flag bit" 402 and determines that "1" applies blind signal separation, and "0" does not apply blind signal separation.

5 is an exemplary diagram illustrating a process of synthesizing data in units of macro blocks in the encoding apparatus according to the present invention.

Here, the encoding apparatus synthesizes two 8 × 8 size blocks into one 8 × 8 size block when synthesizing the image in units of macro blocks.

Data of one macro block 501 in the YUV color area (Y / Cr / Cb) 4: 2: 0 format is internally composed of six 8 × 8 blocks 502. Here, when the two Y component blocks are synthesized into one block (that is, Y1 + Y2 and Y3 + Y4), and the Cb component block and the Cr component block are synthesized into one block (that is, Cb + Cr The macro block 502 is synthesized into a macro block 504 whose data size is compressed in half. At this time, the synthesis 503 sums the pixel values of each position.

As such, the encoding apparatus encodes the macroblock 504 whose data size is compressed in half and transmits the encoded macroblock 504 to the decoding apparatus.

6 is an exemplary diagram illustrating a process of restoring data in units of macro blocks in a decoding apparatus according to the present invention.

In the decoding apparatus, macroblock data, that is, synthesized with half the data size, that is, three 8 × 8 block data 601 are delivered from the encoding apparatus.

Thereafter, the decoding apparatus performs centering to adjust the data average in order to restore the original 8 size block data having 6 8x8 size (602). In this case, the decoding apparatus performs centering to adjust data average through Equation 4 below.

The decoding apparatus copies the adjusted data to separate the signal, and doubles the copied value to generate six 8 × 8 blocks (603). That is, the decoding apparatus randomly generates virtual data so that the signal has the same dimension as the input data and output data in the blind signal separation 604.

In operation 605, the decoding apparatus predicts the residual image value approximating the original data by applying the blind signal separation 604 to the generated six 8 × 8 block data.

7 is a flowchart illustrating a video encoding method using an independent component analysis based signal separation algorithm according to the present invention.

As shown in FIG. 7, the encoding apparatus according to the present invention performs motion estimation / compensation or intra prediction on a current frame to obtain a residual image block of a macro block (S701).

Thereafter, the encoding apparatus determines whether to apply the blind signal separation algorithm at the time of decoding according to the video compression efficiency (S702). In this case, the encoding apparatus proceeds according to a conventional encoding scheme which does not apply the blind signal separation algorithm or does not apply the blind signal separation algorithm.

Hereinafter, the case where the encoding apparatus applies the blind signal separation algorithm will be described.

The encoding apparatus synthesizes each block constituting the residual image block of the macro block (S703). In particular, the encoding apparatus generates and includes the bit stream described above with reference to FIG. 4, which may determine whether to apply the blind signal separation algorithm to the synthesized macroblock at the time of decoding (S704).

Thereafter, the encoding apparatus performs normal discrete cosine transform, quantization, and entropy coding on the macroblock and transmits the macroblock to the decoding apparatus (S705).

8 is a flowchart illustrating a video decoding method using an independent component analysis based signal separation algorithm according to the present invention.

As shown in FIG. 8, the decoding apparatus according to the present invention performs entropy decoding, inverse quantization, and inverse discrete cosine transform on an encoded macro block (S801).

The decoding apparatus determines whether to apply the blind signal separation algorithm by checking the bit stream from the macro block on which the inverse discrete cosine transform is performed (S802).

Hereinafter, a case in which the blind signal separation algorithm is applied to the decoding apparatus will be described.

When the decoding apparatus applies the blind signal separation algorithm, the decoding apparatus performs centering to adjust data averages for each block constituting the macroblock residual image (S803).

Thereafter, the decoding apparatus copies / converts the residual image of the macro block whose data average is adjusted to generate the residual image of the macro block having the same dimension as before encoding (S804).

Thereafter, the decoding apparatus separates the residual image of the macro block into a macro block approximating the original data according to the blind signal separation algorithm (S805).

The decoding apparatus reconstructs an image by summing the macroblocks through a prediction block through motion compensation or intra prediction (S806).

As described above, the method of the present invention may be implemented as a program and stored in a recording medium (CD-ROM, RAM, ROM, floppy disk, hard disk, magneto-optical disk, etc.) in a computer-readable form. Since this process can be easily implemented by those skilled in the art will not be described in more detail.

The present invention described above is capable of various substitutions, modifications, and changes without departing from the technical spirit of the present invention for those skilled in the art to which the present invention pertains. It is not limited by the drawings.

The present invention as described above has the effect of maintaining the proper picture quality while increasing the efficiency of video compression by applying the blind signal separation to video encoding and decoding.

In addition, the present invention has the effect of considering not only the compression efficiency but also the image quality of the image sequence by applying to all macro blocks or blocks of the previous video sequence collectively or by calculating and selectively applying an error for blind signal separation.

In addition, the present invention has an effect of reconstructing an image by referring to a flag value of a bit stream by setting a flag in a bit stream whether implicit signal separation is applied in the encoding device.

Claims (10)

Motion estimation means and intra prediction means to which a current frame is transmitted, motion compensation means to which a reconstructed frame is delivered, discrete cosine transform means, quantization means, reorder means, entropy coding means, inverse quantization means connected to the quantization means, A video encoding apparatus comprising inverse discrete cosine transform means, Block synthesizing means for synthesizing the residual image blocks of the macroblocks obtained by using any one of the motion compensation means and the intra prediction means for each block in the current frame to the discrete cosine transform means; And Adjusting, copying / converting data averages in the macroblocks synthesized by the block synthesizing means and transmitted from the inverse discrete cosine transform means, and restoring the data according to a blind signal separation algorithm, and providing the intra prediction means and the motion compensation means. Blind signal separation means Video encoding apparatus using a signal separation algorithm based on independent component analysis comprising a. According to claim 1, Determining whether to transfer a residual image block of a macro block obtained by using one of the motion compensation means and the intra prediction means to one of the block synthesizing means and the discrete cosine transform means according to video compression efficiency in a current frame Switching control means; And Bit stream generation means for generating a bit stream which can confirm at the time of decoding whether to apply the blind signal separation algorithm according to the determination of the switching control means. Video encoding apparatus using a signal separation algorithm based on the independent component analysis further comprising. According to claim 1, In the block synthesizing means, A video encoding apparatus using an independent component analysis-based signal separation algorithm, characterized by summing and synthesizing pixel values of positions of residual image blocks of a macro block. The method of claim 2, In the bit stream generating means, A video encoding apparatus using an independent component analysis based signal separation algorithm, characterized in that it generates a bit stream indicating whether the blind signal separation algorithm is applied in units of GOP (Group Of Picture) and macro blocks. In the video encoding method which sequentially performs motion estimation, motion compensation, intra prediction, discrete cosine transform, quantization, reorder and entropy coding on the current frame, A first step of determining whether to apply a blind signal separation algorithm upon decoding according to video compression efficiency when the residual image block of the macro block is obtained by performing any one of motion estimation / compensation and intra prediction on a current frame; A second step of synthesizing each block constituting the residual image block of the macro block when the application of the blind signal separation algorithm is determined during decoding; And A third step of generating and including a bit stream capable of determining whether to apply a blind signal separation algorithm to the synthesized macro block in decoding; Video encoding method using an independent component analysis-based signal separation algorithm comprising a. The method of claim 5, After the second step, Performing data compensation and intra prediction after adjusting and copying / converting the data average for each residual image block of the synthesized macro block according to a blind signal separation algorithm. Video encoding method using a signal separation algorithm based on the independent component analysis further comprising. A video decoding apparatus comprising entropy decoding means, reorder means, inverse quantization means, inverse discrete cosine transform means, and motion compensation means and intra prediction means, to which an encoded macroblock is transmitted, Centering means for adjusting an average of data in the residual image of the macro block transferred by the inverse discrete cosine transforming means; Copying / converting means for copying and converting the residual image of the macroblock adjusted by the centering means to generate the residual image of the macroblock in the same dimension as before encoding; And Image separation means for separating the residual image of the macro block generated by the copying / converting means into a macro block approximating the original data according to the blind signal separation algorithm Video decoding apparatus using a signal separation algorithm based on independent component analysis comprising a. The method of claim 7, wherein Switching control means for determining whether to apply a blind signal separation algorithm by checking the bit stream included in the residual image of the encoded macro block; And Switch means for transmitting the residual image of the macro block encoded according to the determination of the switching control means to the image separation means. Video decoding apparatus using a signal separation algorithm based on the independent component analysis further comprising. A decoding method for performing entropy decoding, reordering, inverse quantization, inverse discrete cosine transformation, and intra prediction or motion compensation on a residual image of an encoded macro block, Adjusting a data average for each block from the residual image of the macroblock on which the inverse discrete cosine transform is performed; A second step of generating a residual image of a macro block having the same dimension as before encoding by copying / converting the residual image of the macro block whose data average is adjusted; And A third step of dividing the residual image of the macro block into a macro block approximating original data according to a blind signal separation algorithm; Video decoding method using an independent component analysis based signal separation algorithm comprising a. The method of claim 9, In the third step, Determining whether a blind signal separation algorithm is applied using the bit stream included in the macro block; Video decoding method using a signal separation algorithm based on the independent component analysis further comprising.

Images