KR100586103B1 - Method for moving picture coding - Google Patents

Abstract

The present invention discloses a video encoding method in which the amount of computation is reduced by selecting whether to perform intra prediction on a macroblock based on intra prediction results of blocks when performing encoding for intra prediction in a video encoder. According to an aspect of the present invention, there is provided a method of performing intra prediction mode (0 to 8 prediction mode) encoding on 4 * 4 blocks constituting a macro block for intra mode encoding on an input image; Determining a mode having the best intra prediction coding efficiency for the 4 * 4 blocks subjected to the encoding; Determining the number of 3 to 8 modes among intra prediction modes determined for the 4 * 4 blocks constituting the macroblock; In the intra prediction modes determined for the 4 * 4 blocks constituting the macroblock, when 3 to 8 modes are less than 3, encoding for intra prediction for the macroblock is performed. And performing intra prediction encoding by selecting the most efficient mode among the prediction modes determined in the 4 * 4 blocks without performing encoding for intra prediction.

Video, encoding, block, intra, prediction, macroblock

Description

Video encoding method {METHOD FOR MOVING PICTURE CODING}

1 is a diagram illustrating a structure of a video encoder for encoding a video signal according to the prior art.

FIG. 2 is a diagram for explaining nine intra prediction coding methods for a 4 * 4 block according to the related art. FIG.

3 is a diagram for explaining four intra prediction coding methods for a macroblock according to the prior art;

4A and 4B are diagrams for explaining a method of determining whether to perform an intra prediction encoding process on a macro block according to the present invention;

5 is a flowchart illustrating a method of selecting an intra prediction mode for intra prediction encoding according to the present invention.

* Description of the symbols for the main parts of the drawings *

1: DCT 3: Quantizer

5: Inverse quantization department 7: Inverse DCT

8: Image Memory 9: Motion Predictor

10: VLC

The present invention relates to a video encoding method that, more specifically, can reduce computation amount and reduce power consumption when performing intra prediction encoding on an image.

The current and future communication environment is rapidly changing beyond the distinction between wired and wireless areas and regional countries. In particular, future communication environments, such as IMT-2000, require users as well as video and audio. The trend is to build an environment that provides various information in real time or comprehensively.

In addition, the development of the personal mobile communication system is not only a voice communication in the cellular phone or PCS, but also to transmit text information as well as to access the Internet wirelessly using a personal mobile communication terminal, or to watch videos on TV It is being developed to transmit the data.

In particular, a digital television system that processes a video into digital data, transmits the data in real time, receives the video, and displays the video, and the video transmitted in real time is becoming an essential element in a personal portable terminal using the IMT2000.

In the related art, the portable terminal transmits and receives only human voice. However, due to the development of multimedia and the development of digital information processing technology, various information such as voice and video can be transmitted.

Above all, this technology has been commercialized. First of all, the analog video signal is subjected to special digital processing such as quantization and variable length encoding, and then included in the digital information and transmitted. It greatly contributed to the development of video compression technology that allows abundant information transmission and reception.

Recently, with the development of digital signal processing technology, methods of compressing and transmitting a large amount of video information through a limited bandwidth transmission channel have been developed. However, if an error occurs on the transmission channel, the quality of the reconstructed image is greatly reduced. Occurs.

At this time, error concealment techniques have been studied to recover near to the original image by supplementing the lost information from the neighboring information that is normally restored without using an error correction code to maximize the limited bandwidth.

In particular, in the case of an MPEG (Moving Picture Experts Group), if an error occurs, all information until the next sync signal is found is lost. In addition, since the motion compensation encoding technique is used, the influence of the damaged portion is continued over the subsequent frames.

The present invention is an IMT-2000 video terminal that applies an optimized motion decision value and an intra mode compression algorithm to a hardware accelerator structure that enables fast encoding and quality improvement of video data. Motion decision value of MPEG compression method, which is generally used for wireless image communication, is the most important because it has the largest amount of computation and has a direct influence on the image quality and the amount of bits.

Therefore, many studies on motion judgments have been made continuously, and recently, the motion judgments have the same image quality and bit rate as the full search using the diamond search pattern, and are most commonly used. A three step search method is used.

Various techniques and standards for compressing video or still images have been proposed. The standards for compressing video and still images have the ultimate purpose of compressing and encoding video with more efficient and higher performance.

Currently, MPEG-1, MPEG-2, MPEG-4, and H.261, H.263, H.26L, etc. have been proposed or proposed (adopted) as standards related to video encoding of moving pictures and still images.

Like many international standards, the H.26L standard performs motion estimation and encodes the residual signals resulting from spatio-temporal transform and zigzag scanning. This may be based on encoding in a quantization process.

In such image encoding, an image is encoded on a block basis. For example, a time / spatial transform (for example, a DCT transform, etc.) is performed on an INPUT, and the transform coefficients are quantized and quantized. The compression coefficients are outputted by performing the variable length coding (VLC) by scanning the transform coefficients using a suitable technique such as single scan or double scan.

As described above, when compressing and encoding a moving picture or a still image through block-based motion estimation, each block in an image frame has one motion vector, and there is a degree of correlation between motion vectors of adjacent blocks. .

Therefore, in order to encode a motion vector of each block, a motion vector prediction vector for a block to be currently encoded is obtained using the motion vectors of blocks of adjacent previous frames, and the difference between the motion vector of the adjacent blocks and the motion vector of the adjacent blocks ( Difference Vector encoding is used as a more efficient encoding technique in block-based image encoding.

1 is a diagram illustrating a structure of a video encoder for encoding a video signal according to the prior art.

As shown in FIG. 1, when a video signal photographing a real object is received, a digital conversion is performed, and then an encoding operation is performed to compress the video signal. The digitally converted video signal is subjected to discrete cosine transform in DCT (Discrete Cosine Transform 1), and the image signal converted in frequency form is quantized in quantization unit 3 and then variable in VLC (Variable Length Coding 10). It is transmitted to the outside through the field encoding (entropy) process.

In this case, when the encoded image is an inter mode, a motion vector capable of predicting a previous image is generated from a motion predictor 9. The decoded image frame is reconstructed by the inverse quantization unit 5 and the inverse DCT 7 and stored in the image memory 8. The image stored in the image memory is a reference for predicting motion of an image to be encoded later. Use it as a video.

The first input image is determined as an intra image frame, and intra mode encoding is performed to predict and encode an image in the image frame.

When a next input video is defined as a P picture as an inter mode, motion prediction and compensation are performed on a block of each video frame, and then the error signal is encoded.

In particular, the H.26L international standard is divided into intra coding method using spatial similarity and inter coding method using video frame similarity according to temporal difference.

Although encoding is performed on 8 * 8 blocks in macroblock units consisting of 16 * 16 pixels such as MPEG4 and H.263 +, which are international standards for video encoding, H.26L uses 16 * 16 macroblocks and 4 * 4 blocks. Intra and inter coding of blocks is performed.

The intra encoding method is a method of increasing efficiency by obtaining a difference value between a prediction value obtained due to the directionality of macro blocks and 4 * 4 blocks due to the directionality of adjacent pixels.

In 4 * 4 blocks, encoding is performed by 9 different methods in H.26L, and encoding is performed by 4 methods in units of 16 * 16 macroblocks. Select and proceed with encoding.

 The video bitstream encoded by the encoder is transmitted to the decoder and decoded through the same procedure as that encoded by the encoder.

Although shown in the figure, 2a and 2b, which are not described, represent adders.

FIG. 2 is a diagram for describing nine intra prediction coding methods for a 4 * 4 block according to the related art.

As shown in FIG. 2, intra prediction encoding is performed in a predetermined direction with respect to pixels of 4 * 4 blocks forming a 16 * 16 macroblock.

In addition to the different directions of intra prediction encoding for each pixel, the encoding is performed in H.26L by nine different calculation methods for encoding.

First, in the first method (intra prediction mode 0), a macro block is composed of 16 * 16 blocks, and is divided into four 8 * 8 blocks, which are further divided into four 4 * 4 blocks. In this case, the average value of the neighboring pixels adjacent to the edge of the 8 * 8 blocks is used as a predictive encoding value.

The second method (intra prediction mode 1) calculates an intra prediction coded value by obtaining an average value of 4 * 4 blocks in a vertical direction from top to bottom, as shown, and the third method (intra prediction mode 2) is shown in FIG. As described above, an intra prediction coded value is obtained by calculating an average value of a 4 * 4 block in a horizontal direction from left to right.

The fourth method (Intra Prediction Mode 3) calculates an intra prediction coding value by obtaining a constant average value of adjacent pixels from the upper left corner to the lower right corner. The fifth method (Intra Prediction Mode 4) uses the upper right corner and the lower left corner. It is to obtain a coding value that proceeds by obtaining a constant average value of adjacent pixels at both edges.

The sixth method (intra prediction mode 5) is a method of performing intra prediction encoding in a downward direction while having a constant slope between the direction of the second method and the direction of the fourth method.

The seventh method (Intra Prediction Mode 6) is a method of performing intra prediction encoding in a downward direction while having a constant slope between the direction of the second method (Intra prediction mode 1) and the direction of the fifth method (Intra prediction mode 4). to be.

 The eighth method (intra prediction mode 7) is a method of performing intra prediction encoding in the right direction while having a constant slope between the directions of the fifth method (intra prediction mode 4) and the third method (intra prediction mode 2).

The ninth method (intra prediction mode 8) is a method of performing intra prediction encoding in the right direction while having a constant slope between the directions of the fourth method (intra prediction mode 3) and the third method (intra prediction mode 2).

Intra-prediction encoding is performed on 4 * 4 blocks in the same direction as above, and then four methods are performed on 16 * 16 macroblocks, and then intra prediction encoding mode having the most encoding efficiency is selected.

3 is a diagram for describing four intra prediction coding methods for a macroblock according to the prior art.

As shown in FIG. 3, the intra prediction method for the 16 * 16 macroblock first performs intra prediction encoding on the upper 16 pixels of the macro block in the vertical direction.

The second method is to perform intra prediction encoding on the left 16 pixels in the right horizontal direction. The third method is an average value of the pixels on the left 16 and the upper 16 pixels. Intra predictive encoding is performed.

The fourth method is to perform intra prediction encoding on the left 16 pixels and the upper 16 pixels in a predetermined combination.

As described above, encoding is performed by selecting a prediction mode having the most encoding efficiency among nine intra prediction methods for 4 * 4 blocks and four intra prediction methods for macro blocks as shown in FIGS. 2 and 3.

However, as described above, in order to obtain the best coding efficiency intra prediction in one macroblock in the H.26L image codec, nine predictions are performed for 4 * 4 blocks and four predictions for macroblocks. Therefore, since one intra prediction mode is obtained, the computational amount is very large.

This increase in computation is not only difficult to use efficiently in a recent communication terminal for real-time video, but also has a disadvantage in that the efficiency, such as lack of power depending on the processor.

 An object of the present invention is to provide a video encoding method capable of fast encoding processing without degrading encoding performance by reducing intra prediction methods performed in 4 * 4 blocks and macroblocks in the encoding process for intra prediction. .

In order to achieve the above object, a video encoding method according to the present invention,

Performing intra prediction mode (0 to 8 prediction mode) encoding on 4 * 4 blocks constituting a macro block for intra mode encoding on an input image;

Determining a mode having the best intra prediction coding efficiency for the 4 * 4 blocks subjected to the encoding;

Determining the number of 3 to 8 modes among intra prediction modes determined for 4 * 4 blocks constituting the macroblock;
If 3 to 8 modes are less than 3 among intra prediction modes determined for the 4 * 4 blocks constituting the macro block, encoding for intra prediction of the macro block is performed. Not proceeding with encoding for intra prediction; And
Performing intra prediction on the macro block, comparing the result of performing intra prediction on the 4 * 4 blocks, and selecting a prediction mode having the most encoding efficiency to perform encoding. .

delete

delete

According to the present invention, it is possible to select whether to perform intra prediction encoding on a 16 * 16 macroblock based on intra prediction encoding on a 4 * 4 block, based on intra prediction encoding on a 4 * 4 block. There is an advantage of reducing the amount of encoding computation.

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

4A and 4B are diagrams for describing a method of determining whether to perform an intra prediction encoding process on a macro block according to the present invention.

As shown in FIG. 4A, intra prediction encoding is performed on each of 4 * 4 blocks constituting a macro block to indicate a mode having excellent intra prediction efficiency for each block.

As shown in the intra prediction mode, the first method (intra prediction mode 0) method that uses the average value of neighboring pixels adjacent to the edge for 8 * 8 blocks as the predictive coding value, vertically 4 * 4 blocks from top to bottom A second method of obtaining intra prediction coded values by obtaining an average value in one direction (Intra prediction mode 1), and a third method of obtaining intra predictive coded values by obtaining an average value of 4 * 4 blocks in a horizontal direction from left to right. If (Intra Prediction Mode 2) appears in most of the 4 * 4 blocks constituting the macroblock, four intra prediction methods for the macroblock are performed.

Because the intra prediction modes 0 to 2 appear for the 4 * 4 blocks as described above, it is determined that the variation of the image is small and the spatial similarity is high, so the intra prediction mode is performed for the 16 * 16 for the macroblock. .

However, when a plurality of intra prediction modes for the 4 * 4 blocks other than 0 to 2 modes are selected in the macro block, it is determined that the variation range and complexity of the image is large, and thus the intra prediction for the macro block is performed. Even if it is determined that the prediction cannot be further improved than the intra prediction for the 4 * 4 block belonging to the finer prediction, the intra prediction is omitted for the macro block.

According to the encoder, the intra prediction mode is selected based on the selected number. That is, as shown in FIG. 4B, when there are n or more modes other than the intra modes 0 to 2, it is determined that the change width and complexity of the image is large when the range is 3 or more and within 16 or less. Do not perform intra prediction on the macro block.

If the intra prediction mode for each of the blocks is determined by performing the intra prediction on the 4 * 4 block, 16 intra prediction mode values are obtained for the 16 * 16 block constituting the macroblock.

If the intra prediction mode obtained here is three or more modes other than 0 to 2, the intra prediction mode for the macro block applied to the case where the similarity of the image is wide is not performed because the variation of the image is large and complex.

5 is a flowchart illustrating a method of selecting an intra prediction mode for intra prediction encoding according to the present invention.

As shown in FIG. 5, the video encoder first performs an operation of selecting the most efficient intra encoding mode before encoding the intra mode image frame first applied (S501).

Intra-prediction is performed on 9 4 * 4 blocks of a macroblock constituting an intra picture frame by encoding. Next, an intra prediction mode having the highest coding efficiency is selected from nine intra prediction modes for 4 * 4 blocks (S502).

In this case, when the 4 * 4 block having the intra prediction is completed in units of blocks constituting the macro block, 16 intra prediction mode values are selected for each 4 * 4 block.

Therefore, among the 16 intra prediction mode values for the 4 * 4 blocks constituting the macroblock, how many 0 to 2 modes and how many 3 to 8 modes are determined (S503).

When three to eight intra prediction modes are selected from among 16 intra prediction mode values displayed in the macro block, four intra predictions for the macro block are determined as having a large change in image and a large complexity. The encoding is not advanced (S504).

This is because the intra prediction of the 16 * 16 macroblock is a predictive encoding method that proceeds when the similarity of the images is widely distributed and the change of the image is not severe.

Therefore, when there are three or more three or eight modes of intra prediction modes, intra prediction on the macroblock is omitted, and intra prediction modes having the most coding efficiency among intra predictions performed on 4 * 4 blocks are selected and intra is selected. The predictive encoding is performed (S505).

In addition, when less than three of the 3 to 8 intra prediction modes are selected among the 16 intra prediction mode values displayed in the macro block, the image is determined to be a small change in the image and is widely distributed in the macro block. Four intra prediction encodings are performed (S504 and S506).

Therefore, in the present invention, it is possible to determine the coding efficiency that may appear during intra prediction of a macroblock using the results of nine intra prediction encodings for a 4 * 4 block. Because of the selection, the encoding computation amount is reduced, and the computation speed is increased.

As described in detail above, the present invention uses the result of performing nine intra prediction modes for a 4 * 4 block, and then skips or proceeds the intra prediction mode encoding for a macroblock to reduce the encoding computation amount. It has a reduced effect.

In addition, it is possible to improve the encoding speed by reducing the overall encoding calculation amount, and there is an advantage of reducing the power waste used in the calculation.

The present invention is not limited to the above-described embodiments, and various changes can be made by those skilled in the art without departing from the gist of the present invention as claimed in the following claims.

Claims (4)

Performing intra prediction mode (0 to 8 prediction mode) encoding on 4 * 4 blocks constituting a macro block for intra mode encoding on an input image; Determining a mode having the best intra prediction coding efficiency for the 4 * 4 blocks subjected to the encoding; Determining the number of 3 to 8 modes among intra prediction modes determined for 4 * 4 blocks constituting the macroblock; If 3 to 8 modes are less than 3 among intra prediction modes determined for the 4 * 4 blocks constituting the macro block, encoding is performed for intra prediction of the macro block, and if there are 3 or more, Not performing encoding for intra prediction; And Performing intra prediction on the macro block, comparing the result of intra prediction on the 4 * 4 blocks, and selecting a prediction mode having the most encoding efficiency to perform encoding. Video encoding method. The method of claim 1, The video encoding method of claim 4, wherein the encoding is not performed for intra prediction on the macro block when 3 to 8 modes are 3 to 16 among intra prediction modes determined for the 4 * 4 blocks constituting the macro block. delete delete

Images