CN100536573C - Inframe prediction method used for video frequency coding - Google Patents

Inframe prediction method used for video frequency coding Download PDF

Info

Publication number
CN100536573C
CN100536573C CN 200410000666 CN200410000666A CN100536573C CN 100536573 C CN100536573 C CN 100536573C CN 200410000666 CN200410000666 CN 200410000666 CN 200410000666 A CN200410000666 A CN 200410000666A CN 100536573 C CN100536573 C CN 100536573C
Authority
CN
China
Prior art keywords
prediction
block
current block
pixels
adjacent
Prior art date
Application number
CN 200410000666
Other languages
Chinese (zh)
Other versions
CN1535027A (en
Inventor
孔德慧
孙艳丰
尹宝才
岳文颖
楠 张
王雁来
Original Assignee
北京工业大学
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 北京工业大学 filed Critical 北京工业大学
Priority to CN 200410000666 priority Critical patent/CN100536573C/en
Publication of CN1535027A publication Critical patent/CN1535027A/en
Application granted granted Critical
Publication of CN100536573C publication Critical patent/CN100536573C/en

Links

Abstract

一种用于视频编码的帧内预测方法涉及计算机数字视频编码技术领域,其视频编码是通过摄像机得到原始视频流作为输入,通过视频采集卡进入计算机,并采用JVT提供的视频编码技术,由计算机进行处理与运算。 Of an intra prediction method for video coding relates to a computer technical field of digital video coding, a video encoder which is obtained by a camera the original video stream as an input, into the computer via a video capture card, and using JVT video coding techniques provided by the computer and processing operation. 本发明是基于空间预测的帧内预测模式编码图像的方法,着重定义了利用相邻块中已解码像素的样值来计算DC预测模式的运算规则;并主要根据其上、右上、左、左上和左下块是否可用来重新定义DC模式。 The present invention is an intra picture prediction mode encoding spatial prediction based on a method, defined focus value is calculated by using already decoded samples in adjacent blocks of pixels DC prediction mode arithmetic rule; and mainly based thereon, upper right, left, upper left and whether the lower-left block may be used to redefine the DC mode. 具体编码中,多种预测模式可以被重新组合,并按编码的需要对这些模式做重新的模式排序。 In particular encoding, the plurality of prediction modes can be reassembled, according to re-coding pattern needs to be done to sort these modes. 本发明的方法与JVT标准相比,在不增加任何计算复杂度的情况下提了视频编码质量,也可适当的选取某些模式,在不降低任何图像编码性能的基础上减少编码的计算复杂度。 The method of the present invention compared to the JVT standard, without any increase in computational complexity of video encoding quality mention may be appropriately selected in certain modes, to reduce the computational coding without reducing the performance of any image coding complexity degree.

Description

基于方向的DC预测方法及用于视频编码的帧内预测方法 Based on the direction of DC prediction and intra prediction method for video coding

技术领域 FIELD

本发明涉及一种预测技术,特别是一种基于方向的DC预测方法;本发明还涉及一种帧内预测的方法,特别是一种采用前述方法的视频编码的帧内预测的方法,属于计算机数字视频编码技术领域。 The present invention relates to a predictive technique, particularly to a prediction method based on the direction of DC; The present invention further relates to a method of intra prediction, in particular a method of using the intra-prediction video coding method, a computer belonging to digital video coding technology.

背景技术 Background technique

为了在当前有限的传输带宽和存储媒体中传输和存放图像,我们必须对图像进行压縮编码处理。 For a limited transmission bandwidth and the storage medium storing the image transmission and the current, we have an image compression coding process. 在动态图像的压缩编码技术中,编码算法分帧内编码和帧间编码两种情况。 In the moving image compression encoding technique, a sub-coding algorithm intra-coded and inter-coded in both cases. 其中对视频序列中的第一幅图像或景物变换后的第一幅图像,采用帧内变换编码,其它图像采用帧间编码。 Wherein the first image of a video sequence or image of a scene conversion using intraframe transform coding, interframe coding other pictures. 在现有技术中,帧内编码使用空间预测以利用源信号中的空间统计相关性,帧间编码使用基于块的帧间预测以利用时域统计相关性。 In the prior art intra-coded using spatial prediction to take advantage of the source signal space statistical correlation, inter-coded block based inter prediction to exploit the statistical correlation domain. 在具体编码时,规定了图像内基本处理块的预测方式,如果是采用帧间预测,按相应的算法计算当前块的运动向量,得到其预测值;否则是釆用帧内预测,利用当前块的相邻的已重构的像素按相应的帧内预测技术做预测;进而对预测残差进行变换以去除变换块内的空间相关性,再进行量化;最后,使用现有JVT技术的变长编码或算术编码对量化变换系数信息进行编码。 In a specific coding, the prediction mode defines the basic processing blocks within the image, if interframe prediction, the current block is calculated by the motion vector corresponding algorithm, to obtain the prediction value; otherwise preclude the use of an intra prediction, a current block using adjacent pixels do reconstructed predicted by intra prediction techniques appropriate; further prediction residual transform to remove spatial correlation of the transform block, and then quantized; finally, the use of variable length prior art JVT coding or arithmetic coding to encode the quantized transform coefficient information.

目前,由ITU-T和ISO/IEC JTC1两个国际标准化组织联合形成的音视频标准化联合组织JVT推出的视频编码标准是目前国内外十分流行的一种编码标准,它广泛应用于电视图像压縮、多媒体通讯、多媒体计算机、图像数据库和通信等领域。 Currently, audio and video JVT standardization jointly organized by ITU-T and ISO / IEC JTC1 two International Organization for Standardization formed a joint launch of the video coding standard is very popular at home and abroad a coding standard that is widely used in television image compression , multimedia communications, multimedia computer, an image database and communication fields. 在JVT中一个宏块包括一个16x16的亮度样值块和两个对应的色度样值块,用作视频编解码过程的基本处理单元。 In the JVT a macroblock comprising a 16x16 block of luminance samples and chrominance samples corresponding to two blocks, as the basic processing unit of video encoding and decoding process.

在JVT标准提供的5.0版本视频编码标准的帧内预测技术中,亮度或色度 5.0 intra-prediction video coding standard in the JVT standard provides techniques, the luminance or chrominance

样值的预测采用的是基于pxq块的预测结构,其中p表示块的列数,q表示块的 Samples the predicted value is used in the prediction based on the structure of pxq blocks, where p represents the number of columns of the block, q represents the block

行数,它是利用该pxq块上、右上、左、左上和左下边(图3)己经重构的像素 The number of rows of pixels, which is the use of the pxq blocks, upper right, left, upper left and lower left (FIG. 3) has a reconstructed

样值,按照某些预测模式及其计算规则来预测当前块的像素样值,其中,i=0, 1, ...,2q-l,表示像素行坐标,j=0, 1, ...,2p-l,表示像素列坐标,t.表示当前块左 Samples, and a prediction mode in accordance with certain rules to calculate the predicted pixel sample values ​​of the current block, wherein, i = 0, 1, ..., 2q-l, the ordinate represents the pixel row, j = 0, 1, .. ., 2p-l, it represents a pixel column coordinate, t. the left represents the current block

边一列第i行像素样值,。 An i-th row edge pixel sample values. .表示当前块上边一行第j列像素样值,^表示当前块 Indicates the current block top line j-th column pixel sample values ​​of a current block indicates ^

4第H亍j列的像素样值;其中, H right foot 4 of the pixel sample values ​​of the j-th column; wherein,

在处理亮度样值时,JVT标准定义当块为4x4块、4x8块、8x4块和8x8块时,即p二4或p-8时,c^4或q二8时,采用9种预测模式,这些预测模式和顺序为: When processed luminance samples, the JVT standard definition when the block is 4x4 blocks, 4x8 blocks, when the 8x4 block and 8x8 blocks, i.e. p = 4 or 8 p-time, c ^ 4 or q = 8, the use of 9 prediction modes these modes and prediction order:

模式0:垂直预领!j (vertical prediction) 模式1:水平予页领!j (horizontal prediction) 模式2: DC预测(DC prediction) 模式3: 45度方向预测(diagonal down/left prediction) 模式4: 135度方向预领!j (diagonal down/right prediction) 模式5:112.5度方向预领ij(vertical- right prediction) 模式6: 157.5度方向预测(horizontal-down prediction) 模式7: 67.5度方向预测(vertical- left prediction) 模式8: 22.5度方向预测(horizontal-up prediction) Mode 0: vertical pre collar j (vertical prediction) mode 1:! Level I p collar j (horizontal prediction) Mode 2:! DC prediction (DC prediction) Mode 3: 45-degree direction prediction (diagonal down / left prediction) mode 4 : pre-neck 135-degree direction j (diagonal down / right prediction) mode 5:! 112.5 pre direction collar ij (vertical- right prediction) mode 6: 157.5 prediction direction (horizontal-down prediction) mode 7: 67.5 degree direction prediction (vertical- left prediction) mode 8: 22.5 prediction direction (horizontal-up prediction)

其中,除了DC预测模式,剩余的8种预测模式称为方向预测模式,图4 中的数字标明了各个方向预测模式的指向,未标出的2表示DC预测模式。 Wherein, in addition to DC prediction mode, prediction modes remaining eight directional prediction modes is called, the numbers in FIG. 4 marked prediction modes in various directions, not labeled 2 indicates a DC prediction mode. 其中的DC预测模式定义为: Wherein the DC prediction mode is defined as:

i. 如果<formula>formula see original document page 5</formula>都可用,则所 i. If <formula> formula see original document page 5 </ formula> are available, then the

有预测样本^等于(<formula>formula see original document page 5</formula> There prediction samples is equal to ^ (<formula> formula see original document page 5 </ formula>

ii. 如果<formula>formula see original document page 5</formula>不可用<formula>formula see original document page 5</formula>)可用, ii. If <formula> formula see original document page 5 </ formula> unavailable <formula> formula see original document page 5 </ formula>) is available,

则所有预测样本S,j等于(<formula>formula see original document page 5</formula> iii. 如果<formula>formula see original document page 5</formula>不可用,<formula>formula see original document page 5</formula>可用, 则所有预测样本^等于<formula>formula see original document page 5</formula> All prediction sample S, j is equal to (<formula> formula see original document page 5 </ formula> iii. If <formula> formula see original document page 5 </ formula> unavailable, <formula> formula see original document page 5 </ formula> available, all prediction samples equal to ^ <formula> formula see original document page 5 </ formula>

iv. 如果<formula>formula see original document page 5</formula>都不可用,贝U所有预测样本^等于128, i=0, 1, 2, ...,ql表示像素行坐标,j=0, 1, 2,…,pl, iv. If <formula> formula see original document page 5 </ formula> are not available, all shellfish U ^ prediction samples is equal to 128, i = 0, 1, 2, ..., ql coordinate represents a pixel row, j = 0 , 1, 2, ..., pl,

表示像素列坐标; Column coordinate represents a pixel;

在处理亮度样值时,JVT标准还定义p=q=16时采用4种预测模式,这些预测模式和顺序为: When processed luminance samples, the JVT standard defines p = q = 16 when using 4 kinds of prediction modes, the prediction mode, and the order of:

模式0:垂直予页领lj (vertical prediction) 丰莫式1:水平预须!j (horizontal prediction) 模式2: DC预领!J (DC prediction) 模式3:平板预领!j (plane prediction) Mode 0: vertical to the page collar lj (vertical prediction) Fung Mo formula 1: horizontal pre shall j (horizontal prediction) Mode 2:! DC pre collar J (DC prediction) mode 3:! Tablet Pre collar j (plane prediction)!

其中的DC预测模式定义与亮度块为4x4块、4x8块、8x4块和8x8块的预测模式定义一致。 Wherein the DC prediction mode is defined for the luminance block 4x4 blocks, 4x8 blocks is consistent, the prediction mode and the 8x4 block 8x8 block definition.

在处理色度样值时,JVT标准定义8x8块的4种预测模式和顺序为: 模式0: DC预U (DC prediction) 丰莫式1: 7K平予页须!J (horizontal prediction) 模式2:垂直予页领!j (vertical prediction) 模式3:平板预领!l (plane prediction) When processing the chrominance samples, JVT standard defines four kinds of prediction modes and the order of 8x8 blocks: Mode 0: DC pre-U (DC prediction) Fung Mohs 1: 7K page level shall be J (horizontal prediction) mode 2! : page vertical collar I j (vertical prediction) mode 3:! tablet pre collar l (plane prediction)!

其中的DC预测模式定义与亮度块为4x4块、4x8块、8x4块和8x8块的预 Wherein the DC prediction mode is defined for the luminance block 4x4 blocks, 4x8 blocks pre, 8x4 and 8x8 block of the block

测模式定义一致。 Measurement mode consistent definition.

JVT的预测结构很细,但其在DC预测模式上的预测精准性不够高,并且其对样值的预测模式比较多,如其对亮度样值的预测在4x4块、4x8块、8x4块和8x8块块的预测模式就有9种,导致整个算法的复杂度很高。 JVT predicted structure is very small, but the predictive accuracy in the DC prediction mode is not high enough, and which prediction mode sample values ​​more, as is the prediction of the luminance sample values ​​in the 4x4 block, 4x8 blocks, 8x4 blocks and the 8x8 departments and regions have nine kinds of prediction modes, resulting in a high degree of complexity of the algorithm. 发明内容 SUMMARY

本发明的目的在于为克服现有帧内预测方法的技术缺陷,而提供一种用于视频编码的帧内预测方法。 Object of the present invention is to overcome the disadvantages of the prior art method of intra prediction, intra prediction is provided a method for video coding. 该方法采用前述基于方向的DC预测方法,使得编码过程中帧内预测精度提高,同时降低计算的复杂度。 The method uses the method based on the direction of the DC prediction, the intra prediction encoding process so that accuracy is improved, while reducing the computational complexity.

为了实现本发明的目的,釆用了如下的技术方案: To achieve the object of the present invention, preclude the use of the technical solution is as follows:

6对于帧内预测的任一具有16x16像素的宏块;按照从左至右,从上到下的 6 for either one of the intra prediction macro block having 16x16 pixels; order from left to right, top to bottom

顺序分成若干个具有pxq像素的块;其中,p为任一块的列数,其取值为4、 8 或者16, q为任一块的行数,其取值为4、 8或者16;逐一取出其中任一块作为当前块,进行预测处理;所述预测处理包括对当前块的亮度块进行方向预测和DC预测,对当前块的色度块进行DC预测,并且进行方向预测和/或平板预测; Sequence is divided into several blocks having pxq pixels; wherein, P is any one of the number of columns, whose value is 4, 8 or 16, q is any one of a number of lines, the value of 4, 8 or 16; taken individually either one as the current block, prediction processing; said process comprising luminance prediction block of the current block is DC prediction and prediction direction, the chroma DC block of the current block is predicted, and predict the direction and / or a tablet prediction;

其特征在于,所述的DC预测是一种基于方向的DC预测,包括: Wherein said DC prediction is a DC prediction based on the direction, comprising:

当与当前块相邻的上块、左块均可用时,对当前块任一像素,根据上块中与当前块相邻的同列像素及同列像素的左、右相邻像素,按照对应的权重滤波值进行预测,获得当前块任一像素竖直方向预测值;并且根据左块中与当前块相邻的同行像素及同行像素的上、下相邻像素,按照对应的权重滤波值进行预测,获得当前块任一像素的横平方向预测值;计算所述竖直方向预测值和横平方向预测值的均值,获得DC预测模式下当前块任一像素的预测值; When the current block adjacent to the upper block, a left block may be used for the current block of any one of pixels, according to the upper block to the current block adjacent to the same column of pixels and the same column of pixels of the left and right adjacent pixels, a weight in a corresponding weight filtering prediction value, the pixels in the vertical direction to obtain a prediction value of the current block according to any; the upper left block and the current block pixels adjacent peers and peer pixels, the adjacent pixels, according to the prediction filter corresponding to a weight value, obtain the current predicted value of a block of any horizontal pixels in the horizontal direction; calculating a predicted value of the average predicted value and the vertical-horizontal direction, any of the current block to obtain a prediction value of a pixel in a DC prediction mode;

否则,当与当前块相邻的上块可用时,对当前块任一像素,根据相邻的上块中与所述当前块相邻的同列像素及该同列像素的左、右相邻像素,按照对应的权重滤波值进行预测,获得DC预测模式下当前块任一像素的预测值; Otherwise, when the current block and adjacent upper block is available, the current block according to any one of the left pixel, with the columns of pixels from neighboring blocks adjacent to the current block and the pixels of the same column, and right adjacent pixels, in a corresponding filter weight values ​​to predict the current block to obtain a prediction value of any pixel in the DC prediction mode;

否则,当与当前块相邻的左块可用时,对当前块任一像素,根据相邻的左块中与所述当前块相邻的同行像素及该同行像素的上、下相邻像素,按照对应的权重滤波值进行预测,获得DC预测模式下当前块任一像素的预测值; Otherwise, the current block when the left adjacent block is available, any of the current block of pixels, according to the block adjacent to the left of the current block and the adjacent pixel peer peer pixel, the adjacent pixels, in a corresponding filter weight values ​​to predict the current block to obtain a prediction value of any pixel in the DC prediction mode;

否则,当与当前块相邻的上块、左块均不可用时,设置DC预测模式下当前块任一像素的预测值为128。 Otherwise, when the current block and adjacent upper block, the left block not available, the current block is set to any of the DC prediction mode of a prediction pixel value of 128. 由前述的各个技术方案可知:本发明基于方向的DC预测方法,与现有的相比,由于其在预测之前,采用相邻块进行滤波,使得预测的结果更加精确;另 From the foregoing various technical solutions: a DC prediction direction based on a method of the present invention, as compared with the conventional, prior due to its predicted using neighboring blocks are filtered, so that more accurate prediction results; other

外,本发明用于视频编码的帧内预测方法利用前述的DC预测方法,使得帧内预测更为准确,提高了图像的编码质量;并且,在基于简化的预测模式,在保证 In addition, the present invention is an intra prediction method for video coding of the DC prediction using the aforementioned method, so that more accurate intra prediction, improved encoding quality image; and, based on a simplified prediction mode, to ensure

图像编码性能没有下降的条件下,还能大大减少计算的复杂度。 The image encoding performance under conditions did not decline, but also significantly reduce the computational complexity. 附图说明 BRIEF DESCRIPTION

图1系统框图; 1 a block diagram of the system;

图2当前块与其相邻块位置图; Figure 2 block position adjacent to the current block;

图3pxq块样值的预测结构图; FIG 3pxq prediction block configuration diagram of sample values;

图4 pxq块亮度样值的8个预测方向图; FIG prediction direction in FIG. 8 4 pxq blocks of luminance samples;

图5DC预测模式示意图; FIG 5DC schematic prediction mode;

图6系统流程图; The system flow chart of FIG 6;

图7 8x8块亮度样值的预测结构图; FIG 7 8x8 prediction block configuration diagram of the luminance sample values;

图8亮度样值在9种预测模式下按本发明定义和JVT标准定义下样值信噪比及比特率曲线图; FIG 8 luminance samples at 9 prediction modes sample bit rate and signal to noise ratio graph according to the invention is defined and under the JVT standard definitions;

图9亮度样值在本发明定义的5种预测模式下和JVT标准定义的9种预测模式下样值信噪比及比特率曲线图; 9 samples at 5 brightness prediction modes defined in the present invention, the bit rate and signal to noise sample graph and the 9 prediction modes defined by the JVT standard;

图10色度两个样值中的第一个样值U在本发明定义的DC预测和平板预测2 种预测模式下和JVT标准定义的4种预测模式下样值信噪比及比特率曲线图; 图11色度两个样值中的第二个样值V在本发明定义的DC预测和平板预测2 种预测模式下和JVT标准定义的4种预测模式下样值信噪比及比特率曲线图; 具体实施方式 FIG 10 two chroma sample values ​​of the first sample in the U value defined in the present invention is DC prediction and prediction plate 4 under the two kinds of prediction modes and prediction modes defined by the JVT standard sample and the SNR bitrate curve FIG.; FIG. 11 samples two chroma second samples of the present invention as defined in the V DC prediction and prediction plate 4 under the two kinds of prediction modes and prediction modes defined by the JVT standard bit sample values ​​and SNR FIG curve; DETAILED DESCRIPTION

本发明的系统框图如图1所示,这种用于视频编码的帧内预测方法,是通过视频摄像机得到原始视频序列作为输入,通过视频采集卡后变成视频序列数据进入计算机,并采用JVT提供的视频编码技术,由计算机进行处理与运算。 System block diagram of the present invention is shown in Figure 1, this intra prediction method for video coding, the original video sequence is obtained by a video camera as an input into the video sequence by the video capture card data into the computer, and using JVT video coding techniques provide performs calculation processing by a computer. 其方法步骤为:计算机系统接受采集卡处理好的原始视频流,然后读出接收到的视频序列的一幅图像,并将该图像的像素样值按从左至右,从上到下的顺序 Which method steps: a computer system to accept good acquisition card processing the original video stream, and then read out a video sequence is received, and the pixel values ​​of the image samples in order from left to right, top to bottom

8分成16x16的宏块;从计算机存储器读取到的宏块被送到预测模块,在具体编码时,规定了图像内基本处理块的预测方式,如果是采用帧间预测,按相应的算法计算当前块的运动向量,得到其预测值;否则采用帧内预测,利用当前块 8 is divided into macroblocks of 16x16; are read from the memory is sent to the macroblock prediction module, in the specific coding, the prediction mode defines the basic processing blocks within the image, if interframe prediction is calculated according to the corresponding algorithm the current block motion vector to obtain the prediction value; otherwise, intra-frame prediction, using the current block

的相邻的已重构的像素按相应的帧内预测技术做预测,即按JVT的预测模式和 Adjacent pixels do reconstructed by the corresponding predicted intra prediction techniques, i.e., by the prediction mode and JVT

本发明提出的帧内预测方法进行样值预测,或按本发明提出的简化的预测模式 Proposed intra prediction method of the present invention sample prediction, or simplified prediction mode proposed according to the invention

和预测模式计算方法进行样值预测,然后按JVT标准的方法对预测残差进行变换以去除变换块内的空间相关性,再进行量化;接着,使用现有JVT技术的变长编码或算术编码对量化变换系数信息进行编码,直到本图像编码完成为止, 最后输出本图像编码比特流;读取接收到的序列中的下一幅图像,以此类推直至所有图像均编码完成为止,流程见图6。 Calculation method and a prediction mode predicted sample values, then a method according to the JVT standard prediction residual transform to remove spatial correlation of the transform block, and then quantized; Next, using JVT prior art variable length coding or arithmetic coding the quantized transformation coefficient information is encoded, the image encoding completed until the present, the present final output image encoded bit stream; reading the next image in the sequence received, and so on until all images are encoded is completed, the flow shown in Figure 6.

从计算机存储器读取宏块数据后进入预测模块,如果采用帧内预测,所述对每个选择帧内预测的16xl6宏块中的像素样值进行预测由以下步骤顺次组成: After reading the macroblock data from the computer into the memory module prediction, if the intra-prediction, the predicted values ​​of the pixel samples of each selected intra prediction 16xl6 macroblock in the sequential steps of:

(1) .取一个16x16宏块为当前预测宏块; . (1) take a 16x16 macroblock prediction for the current macroblock;

(2) .将该宏块按从左至右,从上到下的顺序分成pxq , p表示块的列数, 可等于4, 8,或16, q表示块的行数,可等于4, 8,或16; (2). The press is divided into pxq macroblock from left to right, top to bottom, P represents the number of columns of blocks, may be equal to 4, 8 or 16, q denotes the number of line blocks, may be equal to 4, 8, or 16;

(3) .取一个pxq块为当前块; (3) Take a pxq blocks for the current block;

(4) .预测当前块pxq的像素亮度或色度样值; (4) predicting the current pixel block of luminance or chrominance samples of pxq;

(5) .取下一个pxq块作为当前块,重复第(3)至(5)步的过程,直到该宏块预测完毕为止;其中对预测当前块pxq的像素亮度或色度样值,其特征为: (5) Remove a pxq blocks as the current block, repeat (3) to the process of step (5) until completion of the macroblock prediction up; wherein the prediction of the current pixel block of luminance or chrominance samples pxq which characterized by:

1).在处理亮度样值时,当前块为4x4块、4x8块、8x4块和8x8块时,即p:4或p:8时,q-4或q-8时,采用9种预测模式,这些预测模式和顺序为: 1) When processed luminance sample values ​​of the current block of 4x4 blocks, 4x8 blocks, when the 8x4 block, and 8x8 block, i.e., p: 4 or p: 8 When, q-4 or q-8 when using 9 prediction modes these modes and prediction order:

模式o 垂直预测 A vertical prediction mode o

模式l 水平预测 L level prediction model

模式2 DC预测 2 DC prediction mode

模式3 45度方向预测 45-degree direction prediction mode 3

模式4 135度方向预测 135-degree direction prediction mode 4

模式5 112.5度方向预》 Mode 5 112.5 pre direction "

9模式6: 157.5度方向预测模式7: 67.5度方向预测模式8: 22.5度方向预测其中,除了DC预测模式,剩余的8种预测模式称为方向预测模式; 9 Mode 6: 157.5 ° directional prediction modes 7: 8 67.5-degree directional prediction modes: prediction 22.5 degree direction wherein, in addition to DC prediction mode, prediction modes remaining eight directional prediction modes is called;

2) .在处理亮度样值时,当前块为16xl6时,即pq^6时,采用4种预测模式,这些预测模式和顺序为: 2) when processing the luminance samples, 16xl6 the current block, i.e. pq ^ 6, the use of four kinds of prediction modes, the prediction mode, and the order of:

模式0:垂直预测模式1:水平预测模式2: DC预测模式3:平板预测 Mode 0: Vertical prediction Mode 1: Horizontal prediction Mode 2: DC prediction Mode 3: Tablet prediction

3) .在处理色度样值时,当前块为8x8时,即p^q二8时,采用4种预测模式,这些预测模式和顺序为: 3) when processing chroma sample values ​​of the current 8x8 block, i.e. p ^ q = 8, the use of four kinds of prediction modes, the prediction mode, and the order of:

模式0: DC预测模式1:水平预测模式2:垂直预测模式3:平板预测 Mode 0: DC prediction Mode 1: Horizontal prediction Mode 2: a vertical prediction mode 3: Tablet prediction

本发明的特征在于,所述对当前块pxq的像素亮度或色度样值预测时,其中的DC预测模式方法的特征为: Feature of the present invention is that, when the prediction of the current block pxq pixel luminance or chrominance samples, wherein the DC prediction mode method characterized by:

当前块是利用相邻块(U,L,UR,UL, DL)中已解码像素的样值来计算其DC预测模式的,其中,定义符号C表示当前块,符号U表示与当前块相邻的上块, 符号L表示与当前块相邻的左块,符号UL表示与当前块相邻的左上块,符号UR表示与当前块相邻的右上块,符号DL表示当前块表示与当前块相邻的左下块; A current block using neighboring blocks (U, L, UR, UL, DL) already decoded pixel sample values ​​to calculate the DC prediction mode, wherein the symbol C indicates that the current block is defined, the symbol U denotes the current block and the neighboring the upper block, symbol L denotes the left block adjacent to the current block, reference numeral UL represents the upper left block adjacent to the current block, UR represents the current symbol block adjacent to the right block, the current block represented by the symbol DL denotes the current block with adjacent to the lower left block;

1) .与当前块相邻的上块、左块均可用时,定义DC预测模式下当前块所有像素预测值用方向性预测方法求取; 1) adjacent to the current block and the upper block, the left block is available, the predicted pixel value to define all DC prediction mode of a current block is obtained by directional prediction method;

2) .与当前块相邻的上块可用时,定义DC预测模式下当前块所有像素预测值用方向性预测方法求取; 2) adjacent to the current block when the upper block is available, the definition of a DC prediction mode of a current block by obtaining a predicted pixel value for all directional prediction method;

103) .与当前块相邻的左块可用时,定义DC预测模式下当前块所有像素预测 103). When the current block left adjacent block is available, the DC prediction mode is defined for all the pixels of the current block prediction

值用方向性预测方法求取; • Value is obtained by directional prediction method; •

4) .与当前块相邻的上、左块均不可用时,定义DC预测模式下当前块所有像素预测值为128。 4) adjacent to the current block, the left block not available, a DC prediction mode definition of the current block 128 is predicted for all the pixels.

本发明所述一种用于视频编码的帧内预测方法,其特征还在于,在进入帧内预测模块后,所述DC预测模式,可具体采用以下方法定义: The intra prediction method for video encoding according to the present invention, which is characterized in that, after entering the intra-prediction module, and the DC prediction mode, may be defined using the following specific method:

1) 与当前块相邻的上、右上、左、左上和左下面块均可用时,定义DC 预测模式下当前块所有像素预测值用双向预测方法求取,见图5中DC0; 1) adjacent to the current block, upper right, left, upper left, and left following blocks are available, a DC prediction mode definition of the current block using the predicted pixel value for all bidirectional prediction method is obtained, see FIG. 5 DC0;

2) 与当前块相邻的上块可用时,定义DC预测模式下当前块所有像素预测值用竖直方向预测方法求取,见图5中DC1;该方法虽然与垂直预测模式的指向一致,但它们在运算过程中选取的预测像素点的位置、个数和选用的滤波系数不同,在DC预测模式下,当前块像素点用上块中与当前块相邻的相同列像素, 以及该同列像素点左右两相邻像素共三个预测像素点按1, 2, 1权重滤波值进行预测,而垂直预测模式下,当前块像素点用上块中与当前块相邻的相同列像素点直接进行预测; 2) adjacent to the current block when the upper block is available, the definition of a DC prediction mode of a current block by obtaining a prediction value for all pixels in the vertical direction prediction method, see FIG. 5 DCl; Although this method is consistent with the point of a vertical prediction mode, However, the position of the predicted pixel thereof selected in the course of operation, and selection of a different number of filter coefficients in the DC prediction mode, a current block to spend pixel block adjacent to the current block of pixels of the same row, the same column and predicted a total of three pixels around the pixel two pixels adjacent projections by 1, 2, a filter weighting value and the lower vertical prediction mode, a current block to spend the same column of pixels in the pixel block adjacent to the current block directly to predict;

3) 与当前块相邻的左块可用时,定义DC预测模式下当前块所有像素预测值用横平方向预测方法求取,见图5中DC2;该方法虽然与水平预测模式的指向一致,但它们在运算过程中选取的预测像素点的位置、个数和选用的滤波系数不同,在DC预测模式下,当前块像素点用上块中与当前块相邻的相同行像素, 以及该同行像素点左右两相邻像素共三个预测像素点按1, 2, 1权重滤波值进行预测,而水平预测模式下,当前块像素点用上块中与当前块相邻的相同行像素点直接进行预测; 3) adjacent to the current block when the left block is available, the definition of a DC prediction mode of a current block by obtaining a prediction value for all pixels in the horizontal direction horizontal prediction method, see FIG. 5 DC2; Although this method is consistent with the point of a horizontal prediction mode, but their position prediction pixel selected in the course of operation, and selection of a different number of filter coefficients in the DC prediction mode, a current block to spend the same pixel row in the pixel block adjacent to the current block, and pixel peer It predicted a total of three points around the two adjacent pixels for prediction pixels by 1, 2, a filter weighting value and the lower horizontal prediction mode, a current block to spend pixel block adjacent to the current block of pixels in the same row directly prediction;

4) 与当前块相邻的上、左块均不可用时,定义DC预测模式下当前块所有像素预测值为128,这与现有JVT标准相同。 4) adjacent to the current block, the left block not available, a DC prediction mode definition of the current value of all the pixels in the prediction block 128, which is the same current JVT standard.

本发明所述一种用于视频编码的帧内预测方法,其特征还在于,在进入帧内预测模块后,所述DC预测模式,可具体采用以下方法定义数值: The intra prediction method for video encoding according to the present invention, which is characterized in that, after entering the intra-prediction module, and the DC prediction mode, may be specifically defined value by the following method:

(1)首先对当前块已重构出的相邻像素(、~和/,做相应点的低通滤波, (1) First, the current block is reconstructed neighboring pixels (, ~ and / do low-pass filtering corresponding points,

ii放到数组中,记该数组为EP,并记该数组中第m个数组变量为EPm,其中,i=0, 1, ...,2q-l,表示像素行坐标,j=0, 1, ...,2p-l,表示像素列坐标,pxq表示的块大小,p表示块的列数,可等于4, 8,或16, q表示块的行数,可等于4, 8,或16, f,表示当前块左边一列第i行像素样值,^表示当前块上边一行第j列像素 ii into an array, the array is referred to the EP, and note that the array of array variables EPm m, where, i = 0, 1, ..., 2q-l, the ordinate represents the pixel row, j = 0, 1, ..., 2p-l, the column coordinate represents a pixel block size represented pxq, p is the number of columns of blocks, may be equal to 4, 8 or 16, q denotes the number of line blocks, may be equal to 4, 8, or 16, f, representing the current block is a left pixel sample values ​​of the i-th row column, the top line of the current block represents a ^ j-th column of pixels

样值,/表示当前块左上方像素值,a,表示当前块第i行j列的像素样值;m表示数组EP的下标变量; Samples, / represents the top left of the current block pixel value, a, it represents the current pixel block of samples of the i-th row j-th column; m represents an array variable subscript EP;

在以下计算中,符号"》"表示位右移运算; In the following calculations, the symbol " '" represents a bit right shift operation;

EP由下列算法得到: A.如果当前宏块有上边相邻的已重构的像素,即〜.可用,其中,7、0, 1, EP obtained by the following algorithm: A. If the current macroblock has a reconstructed upper adjacent pixels, i.e. ~ available, wherein the 7,0, 1,.

2,…,pl,贝U 2, ..., pl, Tony U

a) EP,二s①; j=0v..,pl a) EP, two s①; j = 0v .., pl

b) 如果当前宏块有右上边相邻的已重构的像素,即〜可用,其中, y=p, p+l, p+2,…,2p-l,贝(J b) If the current macroblock reconstructed pixels adjacent to a right upper side, i.e. ~ available, where, y = p, p + l, p + 2, ..., 2p-l, shellfish (J

EP(1+j+p)=s(p+J); j=0,…,pl EP(i+j+p尸s(p+po ; j二p,…,ql EP (1 + j + p) = s (p + J); j = 0, ..., pl EP (i + j + p dead s (p + po; j two p, ..., ql

否则 otherwise

EP(i+j+p)-EP(p); EP (i + j + p) -EP (p);

j《,…,pl j ", ..., pl

c) EP(】+j+p尸EP(p+j); c) EP (dead + j + p] EP (p + j);

j=p,...,q+l j = p, ..., q + l

d) EP(0)=s0; d) EP (0) = s0;

B.如果当前宏块有左边相邻的已重构的像素,即z,可用,其中, 2,…,ql,贝lj a) EP(—w)-t(i); i=0,…,q-1 B. If the current macroblock has reconstructed pixels adjacent to the left side, i.e., Z, is available, wherein, 2, ..., ql, shellfish lj a) EP (-w) -t (i); i = 0, ... , q-1

b)如果当前宏块有左下边相邻的已重构的像素,g卩々可用,其中, b) If the current macroblock has reconstructed pixels adjacent to the lower left, g Jie 々 available, wherein

12<formula>formula see original document page 13</formula>EP(一i+q尸t(q+i); i=0,…,q-1 EP(+iq尸t(q+qi); i二q,…,pl 12 <formula> formula see original document page 13 </ formula> EP (a i + q dead t (q + i); i = 0, ..., q-1 EP (+ iq dead t (q + qi); i two q, ..., pl

否则 otherwise

EP EP

(小iq尸EP(—q); (Iq small dead EP (-q);

i=0,...,ql i = 0, ..., ql

C) EP(-i—i—q)二EP(-iq); C) EP (-i-i-q) two EP (-iq);

i=p,p+l i = p, p + l

d) EP(o尸t。; d) EP (o dead t .;

C.如果^.可用,并且可用^,其中,…0, 1, 2, 1, 2, ..., pl,贝廿 C. If ^. Is available, and may be ^, wherein, ... 0, 1, 2, 1, 2, ..., pl, shellfish twenty

EP(0)=f; EP (0) = f;

,,ql,其中,y = 0, ,, ql, where, y = 0,

D.定义变量last_pix等于EP(_(p+q)); 取i等于-(p+q),其中,i表示记数器,做以下步骤, D. Definitions variable equal last_pix EP (_ (p + q)); i equals taken - (p + q), where, i denotes Register, do the following steps,

a) .令变量加\¥^^等于(^81:_^^ + (£?(0«1) + £?(1+1) + 2)〉〉2; . A) Let the variable plus \ ¥ ^^ equal to (^ 81: _ ^^ + (£ (0 «1) + £ (1 + 1) + 2??) >> 2;

b) .令变量last_pix等于EP(i); . B) Let the variable equal last_pix EP (i);

c) .令下标为i的数组变量E&等于newjpix; . C) Order array index variable i is equal to E & newjpix;

d) .i增l,转至a),直至i大于(p+q)为止; d) .i by l, go to a), until i is greater than (p + q) so far;

(2) DC预测模式的运算规则 (2) DC prediction mode arithmetic rule

i.如果^,,,都可用,则所有预测样本^等于+巧/+1))»1,见图 i. If ^ ,,, are available, all prediction samples equal to ^ + clever / + 1)) >> 1, see FIG.

5中DCo;其中, 5 DCO; wherein,

i=0, 1, 2,…,ql,表示像素行坐标,j=0, 1, 2, ...,pl,表示像素列坐标; i = 0, 1, 2, ..., ql, coordinate represents a pixel row, j = 0, 1, 2, ..., pl, column coordinate represents a pixel;

13ii.如果f,.不可用,~.可用,则所有预测样本^等于£巧+1,见图5中Dd; 其中,i=0, 1, . 13ii unavailable if f ,. ~ available, all prediction samples equal £ ^ +1 Qiao, Figure 5 Dd;. Wherein, i = 0, 1,

2, ...,ql,表示像素行坐标,j=0, 1, 2,…,pl,表示像素列坐标; 2, ..., ql, coordinate represents a pixel row, j = 0, 1, 2, ..., pl, column coordinate represents a pixel;

iii. 如果^不可用,^可用,则所有预测样本^等于^lw,见图5中DC2; . Iii unavailable if ^ ^ available, all prediction samples is equal to ^ ^ LW, DC2 Figure 5;

其中,i=0, 1, Where, i = 0, 1,

2, ...,ql,表示像素行坐标,j=0, 1, 2, ...,pl,表示像素列坐标; 2, ..., ql, coordinate represents a pixel row, j = 0, 1, 2, ..., pl, column coordinate represents a pixel;

iv. t.都不可用,则所有预测样本Sij等于128,其中,i=0, 1, 2, ...,q画l,表示像素行坐标,j=0, 1, 2,…,pl,表示像素列坐标。 iv. t. are not available, all the prediction samples Sij equals 128, where, i = 0, 1, 2, ..., q L Videos, coordinate represents a pixel row, j = 0, 1, 2, ..., pl , represents the pixel column coordinates.

本发明所述一种用于视频编码的帧内预测方法,其特征还在于,从硬件的实现上考虑,当pxq块为4x4或4x8或8x4或8x8大小时,亮度样值9种预测模式的结构非常复杂,希望用一种更为简单的预测模式来实现图像的压縮,而且又能保证图像压縮的性能不降低。 The intra prediction method for video encoding according to the present invention, which is characterized in that, from the viewpoint of the hardware implementation, when the pxq blocks 4x4 or 4x8 or 8x4 or 8x8 size, a luminance sample values ​​of 9 prediction modes structure is very complex, it is desirable with a more simple prediction model to achieve image compression, image compression and can ensure performance is not reduced. 因此本发明提出可以选取9种预测模式中的部分预测模式,并按编码的需要对这些模式做重新的模式排序;如可采用一种基于5种预测模式的帧内亮度样值预测方法,即DC预测,以及垂直预测、水平预测、45度方向预测、135度方向预测模式; Thus the present invention can be made part of nine kinds of prediction modes selected prediction mode, according to the coding mode needs to be done to re-rank the modes; employed as a luminance sample values ​​of an intra prediction method based on five kinds of prediction modes, i.e., DC prediction and vertical prediction, horizontal prediction, the prediction direction of 45 degrees, 135-degree direction prediction mode;

模式0:垂直予页领!j (vertical prediction) Mode 0: vertical collar pages I j (vertical prediction)!

模式l:水平预测(horizontal prediction) Mode l: horizontal prediction (horizontal prediction)

模式2: DC预测(DC prediction,本发明提出的DC预测模式) Mode 2: DC prediction (DC prediction, DC prediction mode proposed by the present invention)

模式3: 45度方向予页领i (diagonal down/left prediction) Mode 3: 45 degrees to bring the page i (diagonal down / left prediction)

模式4: 135度方向预观!j ( diagonal down/right prediction) Mode 4:! 135-degree direction previewed j (diagonal down / right prediction)

这种简化的预测结构比原来JVT采用的9种预测模式少了4个方向的预测计算,大大减少了计算复杂度。 This simplified structure prediction prediction less than four directions calculated original JVT 9 prediction modes used, greatly reduces the computational complexity.

同样地,对于16x16亮度块和8x8色度块,也可只采用DC预测模式以及垂直预测模式、水平预测模式和平板预测模式中的一种或两种预测模式。 Similarly, for a 16x16 luminance block and 8x8 chrominance blocks, it may be only a DC prediction mode and a vertical prediction mode, horizontal prediction mode and a prediction mode tablet one or both of a prediction mode.

14根据本发明技术方案,按照图1,图6所示,通过视频摄像机得到原始的视频序列作为输入,通过视频采集卡后变成视频数据流进入计算机,并采用JVT 14 According to the present invention, according to FIG. 1, FIG. 6, obtained by a video camera as an input original video sequence, the video capture card into the video data stream into the computer, and using JVT

提供的视频编码技术,对序列中图像的亮度样值做基于8x8块的帧内预测,具 Video coding techniques provide, samples of the luminance image sequence made based on 8x8 block intra prediction, with

体步骤如下: Body following steps:

1. 读取序列中的一幅图像; 1. Read a sequence of images;

2. 将图像按16xl6大小划分成一个个的宏块; 2. Press 16xl6 image size is divided into a number of macro blocks;

3. 取一个16xl6宏块为当前预测宏块; 3. Take a 16xl6 current prediction macroblock macroblock;

4. 将该宏块按从左至右,从上到下的顺序分成8x8的块; 4. The macro block into blocks by 8x8 from left to right, top to bottom;

5. 取一个8x8块为当前块; 5. Take an 8x8 block for the current block;

6. 预测当前块8x8块的像素亮度样值; 6. The prediction of the current pixel block of 8x8 luminance block of samples;

当前的8x8块周边的己经编码完的像素亮度样值位置如图7所示,定义其9 Already encoded 8x8 block surrounding the current pixel luminance finished sample positions 7, 9 define its

种预测模式极其顺序为: Prediction modes is extremely order:

模式0:垂直预测(vertical prediction) 模式1:水平予页领!j (horizontal prediction) 模式2: DC预测(DC prediction) 牛莫式3: 45度方向予页须!j (diagonal down/left prediction) 模式4: 135度方向予页领!j (diagonal down/right prediction) 模式5:112.5度方向预领!j(vertical- right prediction) 模式6: 157.5度方向预领!j(horizontal-down prediction) 模式7: 67.5度方向预测(vertical- left prediction) 模式8: 22.5度方向预测(horizontal-up prediction) Mode 0: vertical prediction (vertical prediction) Mode 1: Horizontal to pages collar j (horizontal prediction) Mode 2:! DC prediction (DC prediction) bovine Mo Formula 3: 45-degree direction to the page to be j (diagonal down / left prediction! ) mode 4: 135-degree direction to bring the page j (diagonal down / right prediction) mode 5:! collar 112.5 pre direction j (vertical- right prediction) mode 6:!! collar 157.5 pre direction j (horizontal-down prediction ) mode 7: 67.5 degree direction prediction (vertical- left prediction) mode 8: 22.5 prediction direction (horizontal-up prediction)

按本发明提出的一种用于视频编码的帧内预测方法,9种模式下本块的预测 The intra prediction method for video coding is proposed according to the invention, the nine modes to the prediction of the present block

模式定义如下: Mode is defined as follows:

1.首先对当前块己重构出的相邻像素t、 ~和/,按JVT方法做相应点的 1. First the current block has the reconstructed neighboring pixels t, ~ and / tap accordingly JVT method

低通滤波,放到数组中,记该数组为EP,并记该数组中第m个数组变量为EPm, 其中,i=0, 1, ...,2q-l,表示像素行坐标,j=0, 1, ...,2p-l,表示像素列坐标,pxq 表示的块大小,p表示块的列数,可等于4, 8,或16, q表示块的行数,可等 Low-pass filtering, placed in the array, the array is referred to EP, and note that the array of array variables EPm m, where, i = 0, 1, ..., 2q-l, the ordinate represents the pixel row, j = 0, 1, ..., 2p-l, the column coordinate represents a pixel block size represented pxq, p is the number of columns of blocks, may be equal to 4, 8 or 16, q represents the number of rows of blocks, and the like may be

15于4, 8,或16, ^表示当前块左边一列第i行像素样值,。 15 at 4, 8, or 16, ^ represents the left of the current block a i-th row pixel sample column. .表示当前块上边一行第j列像素样值,/表示当前块左上方像素值,aij表示当前块第i行j列的像素样值;m表示数组EP的下标变量; It indicates the current block top line j-th column pixel sample values ​​/ value indicates that the current block is the upper left pixel, the pixel of aij represents the current block of samples of the i-th row j-th column; m represents an array variable subscript EP;

在以下计算中,符号"》"表示位右移运算。 In the following calculations, the symbol " '" represents a bit right shift operation.

EP由下列算法得到: EP obtained by the following algorithm:

A. 如果当前宏块有上边相邻的已重构的像素,即。 A. If the current macroblock reconstructed pixels of adjacent upper side, i.e.. .()=0, 1, 2, ..., 7)可用, 则<formula>formula see original document page 16</formula>b) 如果当前宏块有右上边相邻的已重构的像素,即~(乂= 7, 8, ..., 15)可用,贝U . () = 0, 1, 2, ..., 7) is available, <formula> formula see original document page If the current macroblock has a right upper neighboring reconstructed pixels 16 </ formula> b), i.e. ~ (qe = 7, 8, ..., 15) is available, shellfish U

<formula>formula see original document page 16</formula> <Formula> formula see original document page 16 </ formula>

否则 otherwise

<formula>formula see original document page 16</formula> <Formula> formula see original document page 16 </ formula>

B. 如果当前宏块有左边相邻的己重构的像素,即"、0, 1, 2, ..., 7)可用, 则<formula>formula see original document page 16</formula> B. If the current macroblock has been reconstructed left adjacent pixels, i.e., "0, 1, 2, ..., 7) is available, <formula> formula see original document page 16 </ formula>

b) 如果当前宏块有左下边相邻的己重构的像素,即"^8, 9, ..., 15) 可用,则<formula>formula see original document page 16</formula> b) If the current macroblock has been reconstructed lower-left adjacent pixels, i.e., "^ 8, 9, ..., 15) is available, <formula> formula see original document page 16 </ formula>

否则EP(小i,EP(—q); i=0,—,7 Otherwise EP (small i, EP (-q); i = 0, -, 7

c) EP(小,qrEP(-iq); i=8,9 c) EP (small, qrEP (-iq); i = 8,9

d) EP(。尸t。; d) EP (. corpse t .;

C.如果~() = 0, 1, 2,…,7)可用,并且""0, 1, 2,…,7)可用,贝!J C. If ~ () = 0, 1, 2, ..., 7) is available, and "" 0, 1, 2, ..., 7) is available, Tony! J

EP(0)=f; EP (0) = f;

E.定义变量last_pix等于EP(—16); 取i等于-16,其中,i表示记数器,做以下步骤, E. defined variables equal last_pix EP (-16); taken i equals -16, where, i denotes Register, do the following steps,

a) •令变量newjix等于(lastjix + (EP(i)《l) + EP(w) + 2)》2; a) • newjix Let the variable equal to (lastjix + (EP (i) "l) + EP (w) + 2)" 2;

b) .令变量last_pix等于EP①; . B) Let the variable last_pix equal EP①;

c) .令下标为i的数组变量E&等于new_pix; . C) Order array index variable i is equal to E & new_pix;

d) .i增l,转至a),直至i大于16为止; 2.计算各模式下的预测值 d) .i by l, go to a), until i becomes greater than 16; the predicted value is calculated for each mode 2.

a、模式0: 垂直预测(vertical Prediction) a, Mode 0: vertical prediction (vertical Prediction)

使用本模式的必要条件是~(7 = 0, 1, 2, ..., 7)可用,预测样本S,j的产生方法如下: This mode requirement is - (7 = 0, 1, 2, ..., 7) can be used to predict the sample S, j is generated as follows:

aooo — a。 aooo - a. io — a,— aooi = aou = a021 = ^ooo = a012 = 3022 = io - a, - aooi = aou = a021 = ^ ooo = a012 = 3022 =

007 —〜7 — a027 — — ct077 007 -~7 - a027 - - ct077

.—a070 — &0, …=a071 =S1; .-A070 - & 0, ... = a071 = S1;

...=a072 = s2, ... = a072 = s2,

17b、模式1: 水平予页领!l (horizontal prediction) 17b, Mode 1: horizontal page to collar l (horizontal prediction)!

使用本模式的必要条件是""0, 1, 2,…,7)可用,预测样本^的产生方 This mode requirement is " '0, 1, 2, ..., 7) is available, party generating prediction samples ^

法如下: Act as follows:

〜 〜 =…=3 3 = ... = ~ ~

iio = aui 二a a two iio = aui

117 117

ai20 = ai21 二ai22 =…=ai27 = t2 ai20 = ai21 two ai22 = ... = ai27 = t2

ai70 = ai71 = ai72 =.''=ai77 =t7; . Ai70 = ai71 = ai72 = '' = ai77 = t7;

c、模式2:DC预测(DC prediction) c, Mode 2: DC prediction (DC prediction)

土.如果~(7' = 0, 1, 2, ..., 7), ("0, 1, 2,…,7)都可用,则所有预测样本^等于(^_M)+W(/+1))〉〉1; Soil If ~ (7 '= 0, 1, 2, ..., 7), ( "0, 1, 2, ..., 7) are available, all prediction samples is equal to ^ (^ _M) + W (/ +1)) >> 1;

ii.如果f, ("0, 1, 2,…,7)不可用,~(y = 0, 1, 2,…,7)可用,则所有预测样本^等于£巧+1; . Ii If f, ( "0, 1, 2, ..., 7) is not available, ~ (y = 0, 1, 2, ..., 7) is available, the prediction for all samples equal to ^ £ Qiao +1;

化.如果~(乂= 0, 1, 2,…,7)不可用,("0, 1, 2,…,7)可用,则所 Of if - (qe = 0, 1, 2, ..., 7) unavailable, ( "0, 1, 2, ..., 7) is available, then the

有预测样本^等于^V,; ^ Equal samples with a prediction ^ V ,;

iv.如果。 iv. If. .(y、0, 1, 2, ..., 7), ("0, 1, 2, ..., 7)都不可用,则所有预 . (Y, 0, 1, 2, ..., 7), ( "0, 1, 2, ..., 7) are not available, all the pre-

测样本^ij等于128, i=0, 1, 2, ...,7,表示像素行坐标,j=0, 1, 2, ...,7,表示像 ^ Ij equals the measured sample 128, i = 0, 1, 2, ..., 7, the ordinate represents the pixel row, j = 0, 1, 2, ..., 7, as represented by

素列坐标; Prime column coordinate;

其它预测模式和其运算规则与JVT标准相同。 Other prediction mode and the calculation rules same as the JVT standard.

7.确定当前块的最优预测模式 7. Determine the optimal prediction mode of the current block

a. 定义/t表示当前预测模式,令其初始值为0; . A defined / t represents the current prediction mode, to make it an initial value of 0;

b. 由下面的预测残差公式,得到预测模式yt下的预测残差值A^ b. the prediction residuals by the following formula, to obtain a prediction mode in the prediction residual value yt A ^

18A, =a — a' 18A, = a - a '

& kij & Kij

这里,aij表示原始像素亮度样值,^表示模式/t下预测像素亮度样值,^表示预测模式编码号;/=0, 1, 2, ..., ql表示像素行坐标;y=0, 1, 2, ..., p-1 表示像素列坐标; Here, the pixel of aij represents the original luminance samples, schematic ^ / t the prediction pixel luminance samples, ^ represents a number of prediction mode encoding; / = 0, 1, 2, ..., ql coordinate represents a pixel row; y = 0 , 1, 2, ..., p-1 column coordinate represents a pixel;

c. 采用JVT中的编码方法,对各个像素的预测残差做DCT变换(DCT 变换指离散余弦变换)、量化和熵编码,计算当前块在当前模式下的编码比特数;并对各个像素的预测残差做DCT变换和量化后,再进行反量化和反DCT 变换,然后加上预测值Skij,重构块中的各个像素点的亮度样值,记为、j, zt表 . C using the JVT coding method, the prediction residual for each pixel do DCT transform (discrete cosine transform means DCT transform), quantization and entropy coding, calculates the number of coded bits of the current block in the current mode; and each pixel after doing prediction residual DCT transform and quantization then inverse quantization and inverse DCT, and then adds the predicted value SKij, the luminance value of each pixel sample point in the reconstructed block, referred to as a, j, zt table

示预测模式编码号;i'=0, 1, 2,…,ql表示像素行坐标;j'=0, 1, 2,..., p-1表示像素列坐标; It shows a prediction mode code number; i '= 0, 1, 2, ..., ql coordinate represents a pixel row; j' = 0, 1, 2, ..., p-1 column coordinate represents a pixel;

d. 采用JVT中的方法计算该块在当前预测模式下的失真率,记为rdcost; . D calculated using the method of the JVT block distortion ratio at the current prediction mode, referred to as RDcost;

cos, = + /"/nMa x • cos, = + / "/ nMa x •

其中,distortion为当前块的所有像素的原始亮度样值与预测值差的平方禾口, lambda为常数,rate为在当前模式下编码当前块所用比特数; Wherein, distortion of the current block of luminance samples for all pixels of the original and predicted values ​​of the square of the difference between the mouth Wo, lambda is a constant, rate coding a current block in the current mode bits used;

e. M直增l,重复第b, c, d, e步,直至本块所有预测模式都执行一遍为 e. M L linear increase, repeat b, c, d, e step, until all the prediction modes of the present block is executed again

止; stop;

d.比较各个模式下的rdcost,选取rdcost最小的模式为当前最优预测模式; . Rdcost comparison in d each mode, select the smallest current pattern rdcost optimum prediction mode;

8. 将最优预测模式下的预测值作为该块的最后预测值,记为^, z=0, 1, 2,..., ql表示像素行坐标;7=0, 1, 2, ..., pl表示像素列坐标;将最优预测模式 8. The predicted value as the optimal prediction mode of the block is the last predicted value, referred to as ^, z = 0, 1, 2, ..., ql coordinate represents a pixel row; 7 = 0, 1, 2,. .., pl represents a pixel column coordinate; the optimum prediction mode

下的重构值作为该块的最后重构值,记为^', z=0, 1, 2, ..., ql表示像素行坐标;)=0, 1, 2, ..., pl表示像素列坐标; Examples of the reconstruction value of the last block reconstruction value, denoted ^ ', z = 0, 1, 2, ..., ql coordinate represents a pixel row;) = 0, 1, 2, ..., pl column coordinate represents a pixel;

9. 取下一个8x8块作为当前块,重复第6至9步的过程,直到该宏块完全编码为止; 9. Remove a 8x8 block as the current block, repeat steps 6 to 9 procedure, until the complete coding macroblock so far;

19当前预测宏块,重复第3至10步的过程,直到完成整幅图像的编码为止。 19 current prediction macroblock 10 to repeat the process of step 3, until the entire image is encoded so far.

11. 取下一幅图像,重复第2至11步的过程,直到完成整个序列的编码为止。 11. Remove an image, the process is repeated 11 to the second step, until complete encoding the entire sequence.

如果选择简化的预测模式,则每次只需计算第6步中根据需要挑选出的预测模式的预测值,如可只需计算前5种预测模式的预测值。 If a simplified prediction mode selected, the prediction value is calculated only every sixth step selected prediction mode is needed, such as five kinds of predicted values ​​before calculating prediction mode may only. 实例结果 Examples of results

1.利用本发明提出的改进DC预测模式的9种预测模式的预测结构,对一幅1280x720的高清视频序列,以8x8块作为基本处理块,做了10帧全帧内预测测试,并且与现有JVT标准中帧内预测技术在不同量化值下的样值信噪比及比特率做了比较(下表),画出了亮度样值信噪比(记为PSNRY)与比特率(记为Bitrate) 曲线图(图8)。 1. using the prediction structure 9 prediction modes DC prediction mode of the proposed improvements of the present invention, a high-definition video sequences 1280x720 to 8x8 block as a basic processing block 10 made wholly intra prediction tests, and with the current there intra prediction techniques in the JVT standard sample and the signal to noise ratio at different bit quantized values ​​were compared (below), the luminance samples plotted SNR (referred to as PSNRY) and bit rate (referred to as the Bitrate) a graph (FIG. 8).

高清视频序列测试结果:(帧率:30Hz,10帧,1280*720) High-definition video sequence test results :( frame rate: 30Hz, 10 frames 1280 * 720)

qp=29 qp=32 qp=37 qp=43 Gain qp = 29 qp = 32 qp = 37 qp = 43 Gain

JVT PSNRY 40. 74 39. 07 36. 44 33.65 JVT PSNRY 40. 74 39. 07 36. 44 33.65

Bitrate 49975. 51 39797. 21 26065. 39 16003. 66 Bitrate 49975. 51 39797. 21 26065. 39 16003. 66

本发明 PSNRY 40. 74 39. 07 36. 45 33.66 PSNRY present invention 40.74 39.07 36.45 33.66

Bitrate 48273. 29 38174. 86 24659. 59 15053. 78 0.317872 Bitrate 48273. 29 38174. 86 24659. 59 15053. 78 0.317872

从图中可以看出,利用本发明提出的改进DC预测模式的9种预测方法求出的曲线在利用JVT帧内预测方法求出的曲线上方,这表明本发明在没有增加任何复杂度的条件下,图像在压縮性能上的提高。 As can be seen from the figure, the improved DC prediction mode 9 predictive method proposed in the present invention is a curve obtained using JVT upper curve determined intra prediction method, indicating that the present invention does not increase the complexity of any of the conditions , the image to improve the compression performance.

2.利用本发明提出的简化的预测结构,如只取前5种预测模式,保持其原有顺序,分别对一幅1280x720的高清视频序列,做了10帧全帧内测试,与现有JVT标准中帧内预测技术在不同量化值下的样值信噪比及比特率做了比较(下表),画出了样值信噪比与比特率曲线图(图9)。 2. The prediction structure with simplified by the present invention, if only to take the first five prediction mode, to maintain its original order, respectively, a high-definition video sequences of 1280x720, the whole frame 10 made tests, prior JVT intra prediction techniques standard sample and the signal to noise ratio at different bit quantized values ​​were compared (below), to draw the sample graph of signal to noise ratio and the bit rate (FIG. 9).

高清视频序列测试结果:(帧率:30Hz, IO帧,1280*720) High-definition video sequence test results :( frame rate: 30Hz, IO frame 1280 * 720)

qp=29 qp=32 qp:37 qp=43 Gain qp = 29 qp = 32 qp: 37 qp = 43 Gain

20JVT PSNRY 40. 74 39. 07 36. 44 33. 65 20JVT PSNRY 40. 74 39. 07 36. 44 33. 65

Bitrate 49975. 51 39797. 21 26065. 39 16003. 66 Bitrate 49975. 51 39797. 21 26065. 39 16003. 66

本发明 PSNRY 40.71 39. 05 36. 42 33.62 The present invention PSNRY 40.71 39.05 36.42 33.62

Bitrate 49494. 07 39244. 87 25611.6 15811.37 0.070442 Bitrate 49494. 07 39244. 87 25611.6 15811.37 0.070442

从图可以看出,利用本发明提出的简化预测模式方法求出的曲线与利用JVT 帧内预测方法求出的曲线基本重合,这表明本发明在减少了大量的复杂度的条件下,仍能很好的保持图像的压縮性能。 As it can be seen from FIG curve using an intra prediction method using JVT simplified prediction mode of the proposed method of the present invention is determined essentially coincident curve obtained, indicating that the present invention reduces the complexity of a large number of conditions, still maintain a good image compression performance.

3.利用本发明提出的简化的预测结构对两个色度样值进行DC预测(模式0) 和平板预测(模式1)2种预测模式的预测,分别对一幅1280x720的高清视频序列, 做了IO帧全帧内测试,与现有JVT标准中帧内预测技术在不同量化值下的样值信噪比及比特率做了比较(下表),画出了色度样值信噪比(记为PSNRU和PSNRV) 样值信噪比与比特率曲线图(图IO、图ll)。 3 with simplified structure proposed by the present invention, the prediction of the two chrominance samples for DC prediction (mode 0) and flat prediction (mode 1) 2 prediction modes, respectively, of an HD 1280x720 video sequence, do a full frame intra IO test, the JVT standard prior art intra prediction sample and the signal to noise ratio at different bit quantized values ​​were compared (below), the chrominance samples shown SNR (referred to as PSNRU and PSNRV) sample bit rate and SNR graph (FIG IO, FIG. ll).

高清视频序列测试结果:(帧率:30Hz, IO帧,1280*720) High-definition video sequence test results :( frame rate: 30Hz, IO frame 1280 * 720)

<table>table see original document page 21</column></row> <table> <Table> table see original document page 21 </ column> </ row> <table>

从图可以看出,利用本发明提出的简化预测模式方法求出的曲线与利用JVT 帧内预测方法求出的曲线基本重合,这表明本发明在减少了大量的复杂度的条件下,仍能很好的保持图像的压縮性能。 As it can be seen from FIG curve using an intra prediction method using JVT simplified prediction mode of the proposed method of the present invention is determined essentially coincident curve obtained, indicating that the present invention reduces the complexity of a large number of conditions, still maintain a good image compression performance.

Claims (4)

1、一种用于视频编码的帧内预测方法,包括:对于帧内预测的任一具有16×16像素的宏块;按照从左至右,从上到下的顺序分成若干个具有p×q像素的块;其中,p为任一块的列数,其取值为4、8或者16,q为任一块的行数,其取值为4、8或者16;逐一取出其中任一块作为当前块,进行预测处理;所述预测处理包括对当前块的亮度块进行方向预测和DC预测,对当前块的色度块进行DC预测,并且进行方向预测和/或平板预测; 其特征在于,所述的DC预测是一种基于方向的DC预测,包括: 当与当前块相邻的上块、左块均可用时,对当前块任一像素,根据上块中与当前块相邻的同列像素及同列像素的左、右相邻像素,按照对应的权重滤波值进行预测,获得当前块任一像素竖直方向预测值;并且根据左块中与当前块相邻的同行像素及同行像素的上、下相邻像素,按照 An intra prediction method for video coding, comprising: any of an intra prediction for a macroblock having 16 × 16 pixels; order from left to right, top to bottom into a number having p × q block of pixels; wherein, p is any one of a number of columns, the value of 4, 8 or 16, q is a number of rows according to any one, whose value is 4, 8 or 16; taken either individually as a current block prediction process; said process comprising luminance prediction block of the current block is DC prediction and prediction direction, the chroma DC block of the current block is predicted, and predict the direction and / or prediction plate; characterized in that, the DC prediction is described later, based on DC prediction direction, comprising: when the current block and the adjacent block, left block may be used for any of the current block of pixels, according to the block adjacent to the current block in the same column of pixels and columns of pixels with the left and right adjacent pixels, according to the prediction filter weight values ​​corresponding to the right, to obtain any one of a current block of pixels in the vertical direction prediction value; from the left block and the upper block adjacent to the current pixel peers and peer pixels , the neighboring pixels, in accordance with 应的权重滤波值进行预测,获得当前块任一像素的横平方向预测值;计算所述竖直方向预测值和横平方向预测值的均值,获得DC预测模式下当前块任一像素的预测值; 否则,当与当前块相邻的上块可用时,对当前块任一像素,根据相邻的上块中与所述当前块相邻的同列像素及该同列像素的左、右相邻像素,按照对应的权重滤波值进行预测,获得DC预测模式下当前块任一像素的预测值; 否则,当与当前块相邻的左块可用时,对当前块任一像素,根据相邻的左块中与所述当前块相邻的同行像素及该同行像素的上、下相邻像素,按照对应的权重滤波值进行预测,获得DC预测模式下当前块任一像素的预测值; 否则,当与当前块相邻的上块、左块均不可用时,设置DC预测模式下当前块任一像素的预测值为128。 Corresponding weighting filter values ​​predicted, and a current block any transverse horizontal direction prediction value of a pixel; mean of the predicted values ​​is calculated the vertical prediction and horizontal in the horizontal direction, obtaining the prediction value of the current block according to any one pixel in the DC prediction mode; otherwise, when the current block and adjacent upper block is available, the current block according to any one of the left pixel, with the columns of pixels from neighboring blocks adjacent to the current block and the pixels of the same column, and right adjacent pixels, in accordance with the weight corresponding to the predicted filtered values ​​to obtain a prediction value of a current block of pixels of any of the DC prediction mode; otherwise, when the current block left adjacent block is available, any of the current block of pixels, according to a left adjacent block with said upper block adjacent to the current pixel and the peer-peer pixel, the neighboring pixels, in accordance with the predicted value of the corresponding weighting filter to obtain a prediction value of a current block of pixels of any of the DC prediction mode; otherwise, when the upper block adjacent to the current block, when the left block is not available, the current block is set to any of the DC prediction mode of a prediction pixel value of 128.
2、根据权利要求1所述的方法,其特征在于:所述按照对应的权重滤波值进行预测具体是:将相邻的上块中与所述当前块相邻的同列像素及该同列像素的左、右相邻像素按照"1、 2、 1"的权重值滤波后进行预测;或者,将相邻的左块中与所述当前块相邻的同行像素及该同行像素的上、下相邻像素按照"1、 2、 1"的权重值滤波后进行预测。 2. The method according to claim 1, wherein: said filter according to the weight corresponding to the predicted value is specifically: the upper block adjacent to the current block and adjacent pixels in the same column of the same column of pixels left and right neighboring pixels in accordance with the right of "1, 2, 1," the prediction filter weight values; or, on the left adjacent block adjacent to the current block and the pixels of the peers of the peer-pixel, the lower phase o the right pixel in accordance with "1, 2, 1," the prediction filter weight values.
3、 根据权利要求1或2所述的方法,其特征在于:所述对当前块的亮度块进行方向预测和DC预测具体包括:执行对所述当前块亮度块的垂直方向预测、水平方向预测、45度方向预测以及135度方向预测;并且执行各方向预测以及所述DC预测的顺序可任意确定。 3. The method of claim 1 or claim 2, wherein: the performing DC prediction and prediction direction comprises luminance block of the current block: performing vertical luma blocks of the current block prediction, horizontal prediction , 45-degree direction and the 135-degree direction prediction prediction; directions and performs the prediction and DC prediction order can be arbitrarily determined.
4、 根据权利要求1或2所述的方法,其特征在于:所述对当前块的色度块进行DC预测,并且进行方向预测和/或平板预测具体包括:执行对当前块色度块的DC预测,以及执行垂直方向预测、水平方向预测或平板预测之一或其任意组合。 4. The method of claim 1 or claim 2, wherein: said chrominance block of the current block is DC prediction, and performs directional prediction and / or forecast plate comprises: performing current block chroma block DC prediction, and performing vertical prediction, horizontal prediction or predicted one tablet or any combination thereof.
CN 200410000666 2004-01-16 2004-01-16 Inframe prediction method used for video frequency coding CN100536573C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200410000666 CN100536573C (en) 2004-01-16 2004-01-16 Inframe prediction method used for video frequency coding

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200410000666 CN100536573C (en) 2004-01-16 2004-01-16 Inframe prediction method used for video frequency coding

Publications (2)

Publication Number Publication Date
CN1535027A CN1535027A (en) 2004-10-06
CN100536573C true CN100536573C (en) 2009-09-02

Family

ID=34305379

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200410000666 CN100536573C (en) 2004-01-16 2004-01-16 Inframe prediction method used for video frequency coding

Country Status (1)

Country Link
CN (1) CN100536573C (en)

Families Citing this family (45)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2006005659A (en) 2004-06-17 2006-01-05 Canon Inc Device and method for image encoding
JP4501631B2 (en) 2004-10-26 2010-07-14 日本電気株式会社 The image coding apparatus and method, a computer program of the image encoding apparatus, and the mobile terminal
EP1808024B1 (en) * 2004-11-04 2013-04-17 Thomson Licensing Fast intra mode prediction for a video encoder
CN100461867C (en) 2004-12-02 2009-02-11 中国科学院计算技术研究所 Inage predicting encoding method in frame
US7830960B2 (en) * 2005-01-13 2010-11-09 Qualcomm Incorporated Mode selection techniques for intra-prediction video encoding
CN100426868C (en) 2005-01-25 2008-10-15 中国科学院计算技术研究所 Frame image brightness predictive coding method
JP4556124B2 (en) 2005-02-04 2010-10-06 ソニー株式会社 Encoding apparatus and method, a decoding apparatus and method, an image processing system, a recording medium, and program
CN101160973B (en) 2005-04-13 2011-08-17 汤姆逊许可公司 Luma and chroma decoding using a common or three different predictors
CN100518318C (en) 2005-04-13 2009-07-22 华为技术有限公司 Method for enhancing intra-layer and frame prediction
CN103458243B (en) 2005-04-13 2017-03-01 汤姆逊许可公司 Video decoding method and apparatus
GB2425425B (en) 2005-04-22 2011-01-19 Sony Uk Ltd Data processing apparatus
CN100574443C (en) 2005-07-18 2009-12-23 华为技术有限公司 Method for determining predicting mode in increased layer frame and coding and decoding device
CN100505885C (en) 2005-07-18 2009-06-24 华为技术有限公司 Method for determining enhanced layer in-frame prediction mode and encoding and decoding apparatus
CN100397906C (en) 2005-08-24 2008-06-25 天津大学 Fast frame-mode selection of video-frequency information
CN100442857C (en) 2005-10-12 2008-12-10 华为技术有限公司 Method of enhanced layer in-frame predicting method and encoding and decoding apparatus
US9432679B2 (en) 2005-11-01 2016-08-30 Entropic Communications, Llc Data processing system
WO2007093629A1 (en) 2006-02-17 2007-08-23 Thomson Licensing Process for coding images using intra prediction mode
KR100745765B1 (en) 2006-04-13 2007-07-27 삼성전자주식회사 Apparatus and method for intra prediction of an image data, apparatus and method for encoding of an image data, apparatus and method for intra prediction compensation of an image data, apparatus and method for decoding of an image data
CN100515082C (en) 2006-05-23 2009-07-15 中国科学院声学研究所;北京易视威信息技术有限公司 Method for reducing video decoding complexity via decoding quality
KR101011303B1 (en) 2006-07-28 2011-01-28 가부시끼가이샤 도시바 Image encoding and decoding method and apparatus
FR2908007A1 (en) 2006-10-31 2008-05-02 Thomson Licensing Sas Image sequence i.e. video, encoding method for video compression field, involves pre-selecting intra-encoding modes of macro-block with directions that correspond to gradient activities with minimal value to determine encoding mode
KR100856411B1 (en) 2006-12-01 2008-09-04 삼성전자주식회사 Method and apparatus for compensating illumination compensation and method and apparatus for encoding moving picture based on illumination compensation, and method and apparatus for encoding moving picture based on illumination compensation
JP2009044537A (en) 2007-08-09 2009-02-26 Osaka Univ Video stream processing device, its control method, program, and recording medium
CN101115207B (en) 2007-08-30 2010-07-21 上海交通大学 Method and device for implementing interframe forecast based on relativity between future positions
KR101159292B1 (en) 2007-10-15 2012-06-22 니폰덴신뎅와 가부시키가이샤 Image encoding device and decoding device, image encoding method and decoding method, program for the devices and the methods, and recording medium recording program
CN100551075C (en) 2007-10-15 2009-10-14 中兴通讯股份有限公司 Low-complexity in-frame estimation mode selection method
CN100586184C (en) 2008-01-24 2010-01-27 北京工业大学 Infra-frame prediction method
CN100596202C (en) 2008-05-30 2010-03-24 四川虹微技术有限公司 Fast mode selection method in frame
CN101605255B (en) 2008-06-12 2011-05-04 华为技术有限公司 Method and device for encoding and decoding video
CN101677406B (en) * 2008-09-19 2011-04-20 华为技术有限公司 Method and apparatus for video encoding and decoding
CN101389029B (en) 2008-10-21 2012-01-11 北京中星微电子有限公司 Method and apparatus for video image encoding and retrieval
JP5686499B2 (en) * 2009-01-22 2015-03-18 株式会社Nttドコモ Image predictive coding apparatus, method and program, image predictive decoding apparatus, method, and program, as well, encoding and decoding system and method
CN101945270B (en) 2009-07-06 2013-06-19 联发科技(新加坡)私人有限公司 Video coder, method for internal prediction and video data compression
KR101452860B1 (en) * 2009-08-17 2014-10-23 삼성전자주식회사 Method and apparatus for image encoding, and method and apparatus for image decoding
KR20110113561A (en) * 2010-04-09 2011-10-17 한국전자통신연구원 Method and apparatus for intra prediction encoding and decoding using adaptive filter
KR101756442B1 (en) * 2010-11-29 2017-07-11 에스케이텔레콤 주식회사 Video Encoding/Decoding Method and Apparatus for Minimizing Redundancy of Intra Prediction Mode
KR20180117219A (en) * 2010-12-08 2018-10-26 엘지전자 주식회사 Intra prediction method and encoding apparatus and decoding apparatus using same
EP2665274A4 (en) * 2011-01-12 2017-04-12 Mitsubishi Electric Corporation Dynamic image encoding device, dynamic image decoding device, dynamic image encoding method, and dynamic image decoding method
CN102611885B (en) * 2011-01-20 2014-04-30 华为技术有限公司 Encoding and decoding method and device
CN102695061B (en) * 2011-03-20 2015-01-21 华为技术有限公司 Method and apparatus for determining weight factors, and method and apparatus for predicting intra-frame weighting
KR101383775B1 (en) 2011-05-20 2014-04-14 주식회사 케이티 Method And Apparatus For Intra Prediction
TWI642299B (en) * 2011-06-28 2018-11-21 三星電子股份有限公司 Make the video decoding apparatus and coding apparatus and the intra prediction bit stream comprising a non-transitory computer readable medium
CN103096051B (en) 2011-11-04 2017-04-12 华为技术有限公司 A method and apparatus for decoding intra-block signal component image sample points
CN104378644A (en) * 2013-08-16 2015-02-25 上海天荷电子信息有限公司 Fixed-width variable-length pixel sample value string matching strengthened image compression method and device
CN106231303A (en) * 2016-07-22 2016-12-14 上海交通大学 Method for complexity control in HEVC coding by employing prediction mode

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0775109A (en) 1993-08-31 1995-03-17 Victor Co Of Japan Ltd Image information compressing method
CN1129385A (en) 1995-02-13 1996-08-21 大宇电子株式会社 Method and apparatus for encoding a video signal using pixel-by-pixel motion prediction
JPH08340535A (en) 1995-06-14 1996-12-24 Nippon Telegr & Teleph Corp <Ntt> Inter-frame predict method
JPH1175196A (en) 1997-08-27 1999-03-16 Sharp Corp Moving image coder and moving image decoder
JPH1198528A (en) 1997-09-19 1999-04-09 Fujitsu Ltd Method for compressing stereoscopic animation and device therefor
CN1455600A (en) 2003-05-19 2003-11-12 北京工业大学 Interframe predicting method based on adjacent pixel prediction

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH0775109A (en) 1993-08-31 1995-03-17 Victor Co Of Japan Ltd Image information compressing method
CN1129385A (en) 1995-02-13 1996-08-21 大宇电子株式会社 Method and apparatus for encoding a video signal using pixel-by-pixel motion prediction
JPH08340535A (en) 1995-06-14 1996-12-24 Nippon Telegr & Teleph Corp <Ntt> Inter-frame predict method
JPH1175196A (en) 1997-08-27 1999-03-16 Sharp Corp Moving image coder and moving image decoder
JPH1198528A (en) 1997-09-19 1999-04-09 Fujitsu Ltd Method for compressing stereoscopic animation and device therefor
CN1455600A (en) 2003-05-19 2003-11-12 北京工业大学 Interframe predicting method based on adjacent pixel prediction

Also Published As

Publication number Publication date
CN1535027A (en) 2004-10-06

Similar Documents

Publication Publication Date Title
CN1331353C (en) Method for sub-pixel valve interpolation
RU2400941C1 (en) Device for coding of dynamic images, device for decoding of dynamic images, method for coding of dynamic images and method for decoding of dynamic images
CN103118254B (en) Moving picture coding apparatus and the moving picture decoding apparatus
JP5401009B2 (en) Intra prediction coding of an image, decoding method and apparatus
CN102316320B (en) Skip macroblock coding
CN101371584B (en) Method and apparatus for reducing the resolution of the multi-view video encoding update mode
CN100407224C (en) Bit plane coding and decoding for the macroblock field / frame coding type information
CN102084655B (en) Video encoding by filter selection
CN103957406B (en) Video coding of large macroblocks with
JP5583840B2 (en) Intra prediction execution method and apparatus using an adaptive filter
CN1224268C (en) Moving picture coding apparatus and moving picture decoding apparatus
CN100553321C (en) Coding dynamic filters
CN1956546B (en) Image coding apparatus
CN106385583B (en) Image processing apparatus and method
CN1297148C (en) Moving picture encoding/transmission system, moving picture encoding/transmission method, and encoding apparatus, decoding apparatus, encoding method decoding method and program usable for the same
KR101228651B1 (en) Method and apparatus for performing motion estimation
CN100361536C (en) Device and method for predicting and encoding image and recording medium
CN101385347B (en) Method of and apparatus for video intraprediction encoding/decoding
CN102595131B (en) Encoder for encoding video signal data for an image block
CN100534194C (en) Methods and apparatus for video intraprediction encoding and decoding
CN101911702B (en) Method and device for quantization of video module coefficient for CABAC supported video coding process
CN101112103B (en) Method of effectively predicting multi-layer based video frame, and video coding method and apparatus using the same
KR101196429B1 (en) Video transcoding method and apparatus, and motion vector interpolation method
KR101619972B1 (en) Apparatus and method for coding/decoding image selectivly using descrete cosine/sine transtorm
CN101346998B (en) Video encoding method, decoding method, device thereof

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted