CN100417228C - Method of selecting in frame prediction mode based on H.264/AVC standard frame image - Google Patents

Method of selecting in frame prediction mode based on H.264/AVC standard frame image Download PDF

Info

Publication number
CN100417228C
CN100417228C CN 200510086742 CN200510086742A CN100417228C CN 100417228 C CN100417228 C CN 100417228C CN 200510086742 CN200510086742 CN 200510086742 CN 200510086742 A CN200510086742 A CN 200510086742A CN 100417228 C CN100417228 C CN 100417228C
Authority
CN
China
Prior art keywords
prediction mode
value
method
intra prediction
mode
Prior art date
Application number
CN 200510086742
Other languages
Chinese (zh)
Other versions
CN1753500A (en
Inventor
刘昌贺
展 徐
刚 李
睿 梁
民 王
飞 程
郭更生
Original Assignee
连展科技(天津)有限公司
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 连展科技(天津)有限公司 filed Critical 连展科技(天津)有限公司
Priority to CN 200510086742 priority Critical patent/CN100417228C/en
Publication of CN1753500A publication Critical patent/CN1753500A/en
Application granted granted Critical
Publication of CN100417228C publication Critical patent/CN100417228C/en

Links

Abstract

本发明提出一种基于H.264/AVC标准的帧图像的帧内预测模式的选择方法,它包括亮度部分的帧内预测模式选择方法,和色度部分的帧内预测模式选择方法。 The present invention provides a method of selecting an intra prediction mode frame image H.264 / AVC standard based on an intra prediction mode of the luminance which comprises a selected portion of a method, and the intra chrominance prediction mode selection method section. 其中,亮度部分的选择方法的步骤是,先把帧图像划分为多个宏块,设定差异度门限值,计算每个宏块的差异度,比较所得的差异度和门限值,根据不同的比较结果使用不同尺寸的预测模式,然后计算每种预测模式的SATD值,根据所得的值,获得优选的预测模式,对这些优选的预测模式运行率失真优化,最终确定最佳模式。 Wherein the step of selecting a method luminance portion is first frame image into a plurality of macroblocks, difference degree threshold is set, calculates the degree of difference of each macroblock, and comparing the resulting difference degree threshold value, in accordance with Comparative results using a different prediction modes of different sizes, and SATD value is calculated for each prediction mode, based on the value obtained, preferably obtained prediction mode, the prediction mode rate-distortion optimization of these preferred, and the optimum mode. 本发明的帧内预测模式选择方法在没有降低图像质量的前提下,能很大幅降低编码运算量,明显减少视频编码耗时,使得H.264/AVC的视频具有更好的实时实用性。 Intra-forecast mode selecting method of the present invention on the premise of no reduction in image quality, it can significantly reduce the amount of coding operation, significantly reducing the time consuming video coding, such H.264 / AVC video in real time with better practicability.

Description

一种基于H,264/AVC标准的帧图像的帧内预测模式选择方法技术领域本发明涉及视频编码领域,具体地涉及H.264/AVC标准中的帧5 图像的帧内预测模式的选择方法。 Based on H, an intra prediction mode frame images 264 / AVC standard selection TECHNICAL FIELD The present invention relates to video coding, and in particular to a method of selecting an intra prediction mode of the image frame 5 H.264 / AVC standard . 背景技术近两年新出现的视频编解码标准H.264/AVC是由ITUT和MPEG 一起联合开发的新一代视频编解码标准。 BACKGROUND past two years, emerging video codec standard H.264 / AVC and MPEG ITUT together by the joint development of next-generation video codec standard. H.264/AVC项目最初的目标是希望新的编解码器能够在比相对以前的视频标准(比如MPEG-2io 或者H.263 )低很多的码率下(比如说, 一半或者更少)提供很好的视频质量;同时,并不增加很多复杂的编码工具,使得硬件难以实现。 H.264 / AVC initial goal is to program the new codecs (for example, half or less) lower than the opposite previous video standards (such as MPEG-2io or H.263) provide many code rates good video quality; at the same time, does not increase the number of complex coding tools, making it difficult to achieve hardware. 另外一个目标是可适应性,即编解码器能够在一个很厂的范围内使用(比如说,既包含高码率也包含低码率,以及不同的视频分辨率), 并且能在各种网络和系统上(比如组播,DVD存储,RTP/IP包网络,15 ITU-T多媒体电话系统)工作。 Another objective is adaptive, i.e. the codec can be used (for example, comprising both high rate also includes a low bit rate, video resolution and different) over a range of plants, and can be in various network and a system (such as multicast, DVD storage, RTP / IP packet networks, 15 ITU-T multimedia telephony system) working. 为了达到上述目的,许多新的技术首次在该标准中釆用,例如帧内编码的空间预测,可变块大小的运动估计补偿,多个参考帧以及自适应二进制编码,这些技术大大增加了^1.264/八¥(:编码器的运算复杂度。在这些新的特性中,通过运用可变大小块和多个方向预测的技术,极大的丰富了帧内预测模式的数目。20 2002年举行的"ISO/IEC JTC1/SC29/WG11与ITU-T SG16 Q.6,JVT第四次会议,,中发表的文章《比较:H.26L帧内编码与JPEG2000》, 其中的研究结果表明,借助于这些新技术H.264/AVC帧内编码的性能超过了JPEG-2000的静态图像压缩标准。H.264/AVC中采用率失真优化(RDO)的方法来进行帧内预测模式的选择,从而达到更高的编5码效率。然而,现有技术中R264/AVC编码技术是有缺陷的。这是因为, 依据H.264/AVC中率失真优化(RDO)的定义,对视频序列进行编码时,对帧图像必须使用所 To achieve the above object, a number of new technologies for the first time preclude the use of this standard, for example, intra spatial prediction coding, variable block size motion estimation compensation, a plurality of reference frames and adaptive binary coding, these technologies greatly increased ^ 1.264 / oct ¥ (:. encoder computational complexity of these new features, the predicted through the use of variable size block and a plurality of directions technology, greatly enriched the intra-prediction mode number held in 2002 .20 the "ISO / IEC JTC1 / SC29 / WG11 and ITU-T SG16 Q.6, JVT fourth meeting ,, published in the article" Compare: H.26L intra-frame coding and JPEG2000 ", in which the research results show that the aid the performance of these new technologies intraframe coding H.264 / AVC still exceeds the JPEG-2000 image compression standard .H.264 / AVC method using a rate-distortion optimization (RDO) to select an intra prediction mode, whereby 5 yards achieve higher coding efficiency. However, the prior art R264 / AVC coding technique is defective. this is because, according distortion optimization (RDO) definition / of the AVC H.264, encodes the video sequence when, using the frame image to be 有的帧内预测模式各测试一遍,然后找到5 —个从RDO角度来讲最佳的预测模式。具体的说,发表在期刊《视频技术电路与系统的IEEE学报》20(B 年7月刊第1-19页的文章《纵观汪264/八¥(:视频编码标准》中提到, 在H.264/AVC中总共定义了13种亮度部分和4种色度部分的帧内预测模式,它们分别是9种4x4的亮度部分,4种16xl6的亮度部分和io 4种8x8色度部分的帧内预测模式。为了选择最佳的帧内预测模式, R264/AVC要求在编码时,必须对每一帧图像的每一种模式都进行一次RDO运算。而率失真优化(RDO)的方法是通过计算率失真度(RDcost),根据所得的结果选择最佳模式,其中使用的计算公式是: i^eos/二D + i?x/i 。其中,D表示的是对应预测模式的失真度,R15 表示编码之后的码流速率,2表示的是一个拉格朗日系数。也就说对图像的亮度部分总共要进行13种预测模式的RDO计算, Some intra-prediction mode for each test again, then found 5 - Best prediction mode from the perspective of RDO specifically, published in the journal "IEEE Journal of Circuits and Systems Video Technology" 20 (B July issue of 1-19 page article "Throughout Wang 264 / octave ¥ (: video coding standard" is mentioned in the H.264 / AVC defined a total of 13 kinds of brightness part and four intra prediction mode of the chroma part, they are part of the luminance of the nine kinds of 4x4, 4 16xl6 IO luminance portion and four kinds of intra-prediction modes of 8x8 chrominance portion. in order to select the optimal intra prediction mode, R264 / AVC coding requirements, must each frame image of each mode RDO calculation are performed and a rate-distortion optimization (RDO) method by calculating the rate distortion (RDcost), selects an optimum mode based on the results obtained, which is calculated using: i ^ eos / two D + i? x / i. where, D denotes a prediction mode corresponding to the degree of distortion, R15 represents a code stream after the encoding rate, a 2 represents the Lagrange multiplier. also told the luminance portion of the image to be a total of 13 kinds of prediction RDO calculation mode, 图像的色度部分也要进行4种预测模式的RDO计算。尽管率失真优化(RDO ) 的方法可以保证编码后的图像获得最佳的压缩效率和图像质量,但是由于率失真优化的计算复杂度非常高,计算量庞大,它的运算是非常20 占用时间的,因此,目前的H.264/AVC编码技术不适合一些需要实时应用的场合,例如视频会议等。另一方面,在JVT提供的参考编解码模块JM98中,还提供了另外一种模式选择方法,该方法是基于绝对变换差之和(SATD)的计算。 Chrominance portion of the image must also be calculated four kinds of prediction RDO mode. While the rate-distortion optimization (RDO) method ensures optimum coded image compression efficiency and image quality, but due to the complexity of the rate-distortion optimization calculation very high, a large amount of calculation, its operation is very 20 takes time, and therefore, the current H.264 / AVC encoding technology is not suitable for some applications requiring real-time applications, such as videoconferencing. on the other hand, provided in the JVT reference JM98 codec module, there is also provided a method of selecting another mode, the method is based on the calculated sum of absolute transformed difference (the SATD) a. 将帧内预测模式的失真度矩阵进行一次哈达玛(Hadamard)变换, 25 并取变换后矩阵各个元素的绝对值,再把所有的绝对值相加就可以得到与该预测模式对应的SATD的值。 The distortion matrix intra-prediction mode once the Hadamard (the Hadamard) transform, the transform matrix 25 and after taking the absolute value of each element, then all of the absolute values ​​can be obtained corresponding to the value of the prediction mode SATD . 经过实验统计可以发现,基于计算SATD的方法所获得的候选的预测模式和基于计算RDcost的方法所获得的预测模式在很大程度上是一致的。 Statistics can be found through experiments, the prediction mode SATD calculation method based on the obtained candidate and the prediction model is based on the calculation method RDcost obtained are consistent to a large extent. 也就是说由基于率失真优化(RDO)的方法最终所获得的最佳模式有很大的可能性是根据计算SATD的方法所选择的几种优先选择的预测模式中的一个。 That is the optimum mode based on the rate distortion optimization method (RDO) is finally obtained has a strong possibility that the prediction model according to several methods calculate SATD selected in preference. 而相比5 RDO运算,SATD的计算在计算量和运算时间上都有着非常明显的优势。 And compared to 5 RDO, calculates SATD in the calculation of the amount of computing time and have a very distinct advantage. 因此,通过先计算宏块失真度,确定预测模式的尺寸,初步减少候选预测模式的数目,再计算其SATD进一步获得优选的预测模式, 最后对这些模式进行RDO运算,选择最佳模式。 Accordingly, by calculating the first distortion macroblock, a prediction mode determination size, reducing the number of initial candidate prediction mode, and then calculate SATD further preferably obtained prediction mode, the final mode RDO calculation of these, selects an optimum mode. 这样的方法能减少10 RDO运算的次数,以此获得一种新的帧图像的帧内预测模式选择方法,使得这种新的方法相对于现有技术的标准方法,具有在不降低图像品质的基础上,计算量和运算时间上的大幅度节省,是有可能买现的。 Such method can reduce the number of calculations 10 RDO, in order to obtain the intra prediction mode of a new frame image selection method, such that this new method compared to standard method of the prior art, without reducing the image having a quality Based on the substantial savings in computation and calculation time, it is possible to buy now is. 发明内容15 ( — )要解决的技术问题本发明的目的是要克服现有技术的缺陷,提出一种应用于H.264/AVC标准的、能有效的减少运算量和运算时间的视频编码方法,解决现有技术中编码复杂度过高、运算量过大、耗时过多而造成的应用限制的问题,使得H.264/AVC标准的视频编码能更具有实时20 应用性。 Summary of the Invention 15 (-) object of the Problem to be Solved by the present invention is to overcome the drawbacks of the prior art, be applied to provide a H.264 / AVC standard, can effectively reduce the amount of computation and calculation time video encoding method to solve the prior art coding complexity is too high, the operation is too large, time-consuming problems caused by excessive application of limitations, making H.264 / AVC video coding standard real-time 20 more have applied. (二)技术方案本发明提出一种H.264/AVC帧图像亮度部分的帧内预测模式选择方法,它包括以下步骤:(1)将视频帧图像划分为为相应数目的16x16大小的宏块;(2) 设定视频帧图像宏块的差异度门限值;(3) 计算视频帧图像每个宏块的差异度;(4) 比较步骤(3)所得的差异度和步骤(2)所得的门限值;(5) 根据不同的比较结果选择不同尺寸的帧内预测模式,并计5算每个预测模式所对应的绝对变换差之和SATD的值;(6) 根据步骤(5)所得的SATD值,获得优选的预测模式;(7) 对步骤(6)所得的优选的预测模式,通过率失真优化,确定最佳模式.上述的帧图像亮度部分的帧内预测模式选择方法,其中,对于步io 骤(2)的门限值,不是一个固定值,根据不同的编解码图像质量和传输速率的要求来调整门限值的具体值.其中一种优选的方案是,门限值Dfl50。 (Ii) aspect of the present invention provides an intra prediction mode of one kind of H.264 / AVC frame image luminance portion selection method, comprising the steps of: (1) the image into a video frame into a corresponding number of size 16x16 macroblock ; (2) setting the degree of difference video gate frame image macroblock value; (3) calculating the difference for each macroblock of a video frame image; (4) the step of comparing the difference of step (3) obtained in (2) the resulting threshold; and (5) selecting an intra prediction mode different sizes depending on the comparison result, and the value calculated from 5 each prediction mode corresponding to the sum of absolute transformed difference and SATD; and (6) according to step (5 ) resulting SATD value, preferably obtained prediction mode; (7) of step (6) of the prediction mode is preferably obtained by rate-distortion optimization, to determine the best mode of the intra prediction mode of the luminance portion of a frame image selecting method. wherein the threshold value for the io step step (2), not a fixed value, the threshold value is adjusted according to the specific requirements of different image quality and transmission rate codec. one preferred embodiment, the door limit Dfl50. 上述的帧图像亮度部分的帧内预测模式选择方法, 一种优选的方案是,步骤(3)中差异度的计算使用公式<formula>formula see original document page 8</formula>, 其中Lxy表示的是一个宏块中的象素点的亮度值;如果用一个宏块中的象素点的亮度值构造一个矩阵,则差异度Discrepancy就是该矩阵的标准方差。 An intra prediction mode of the luminance portion of a frame image selecting method, a preferred embodiment, step (3) using the difference degree calculation formula <formula> formula see original document page 8 </ formula>, which represents the Lxy is the luminance value of the pixel dots of a macroblock; if the luminance value of pixel dots of a configuration of a macro-block matrix, the difference of the standard deviation is discrepancy matrix. 上述的帧图像亮度部分的帧内预测模式选择方法, 一种优选的方血曰步骤(3)中差异度的计算公式还可以是<formula>formula see original document page 8</formula>,其中Lxy表示的是一个宏块中的象素点的亮度值;这个差异度的值也表示,如果用宏块中的象素点的亮度值构造一个矩阵,则差异度D就是该矩阵的方差。 An intra prediction mode of the luminance portion of a frame image selection method, a preferred formula for calculating said blood side of the step (3) may also be differences in the degree of <formula> formula see original document page 8 </ formula>, wherein Lxy It represents the luminance value of pixel dots of a macro block; this value also represents a degree of difference, a luminance value matrix if the configuration of the pixel dots macroblock, the difference D is the matrix of variance. 案是上述的帧图像亮度部分的帧内预测模式选择方法, 一种优选的方案是,步骤(5)所选择的不同尺寸的帧内预测模式,有16xl6的帧内预测模式,和4x4的帧内预测模式;对于差异度小于门限值的宏块,选用16x16的帧内预测模式,对于差异度大于门限值的宏块, 5选用4x4的帧内预测模式。 Selecting an intra prediction mode is the case of the above-described method of frame image luminance portion, a preferred embodiment, the intra prediction mode in step (5) of the selected different size, there is the intra prediction mode 16xl6, and 4x4 frame intra prediction mode; macroblock for differences less than the threshold value, the selection of an intra 16x16 prediction mode, the difference is greater than the threshold value for the macroblock, the intra 4x4 prediction mode is selected 5. 对于这两种不同尺寸的预测模式,进一步计算每一种预测模式对应的SATD值。 For these two different sizes of prediction modes, further calculates SATD value corresponding to each prediction mode. 上述的幀图像亮度部分的帧内预测模式选择方法, 一种优选的方案是,根据步骤(6)所得的SATD值,对9种4x4亮度预测模式, 取其中SATD值最小的3种模式,作为优选的预测模式;根据步骤(6) io 所得的SATD值,对4种16x16亮度预测模式,取其中SATD值最小的2种模式,作为优选的预测模式。 An intra prediction mode of the luminance portion of a frame image selection method, a preferred embodiment, the step (6) obtained in SATD value, for nine 4x4 luma prediction mode, the minimum SATD value which takes three modes, as preferred prediction mode; io the SATD value obtained in step (6), four kinds of 16x16 luma prediction mode, the minimum SATD value which takes two modes as a preferred mode of prediction. 上述的帧图像亮度部分的帧内预测模式选择方法,其中,步骤(1) 和步骤(2)的操作顺序可以互换.本发明还提出一种H.264/AVC帧图像色度部分的帧内预测模式15选择方法,其特征在于,它包括以下步骤:(1) 将视频帧图像划分为为相应数目的16x 16大小的宏块;(2) 计算视频帧图像每个宏块的SATD值;(3) 根据步骤(2)所得的SATD值,获得优选的预测模式;(4) 对步骤(3)所得的预测模式,通过率失真优化,确定最佳20 模式。 An intra prediction mode of the luminance portion of a frame image selection method, wherein the sequence of operations in step (1) and step (2) can be interchanged. The present invention also provides a frame one kind of H.264 / AVC frame image chrominance portion 15 intra prediction mode selection method, which is characterized in that it comprises the following steps: (1) a video frame image is divided into a number corresponding to the size of 16x 16 macroblock; (2) a video frame image is calculated for each macroblock SATD value ; (3) according to step (2) obtained in SATD value, preferably obtained prediction mode; (4) step (3) prediction mode obtained by rate-distortion optimization, to determine the best mode 20. 上述的帧图像色度部分的帧内预测模式选择方法, 一种优选的方案是,根据步骤(3)所得的SATD值,对4种8x8色度部分帧内预测模式,取其中SATD值最小的2种模式,作为优选的预测模式。 An intra prediction mode of the chrominance portion of the frame image selection method, a preferred embodiment, the step (3) obtained in SATD value, four kinds of 8x8 chrominance portion intra prediction mode, which takes a minimum SATD value 2 modes, as a preferred mode of prediction. 本发明还提出一种H.264/AVC帧图像的模式选择方法,它包括25 上述的H.264/AVC帧图像亮度部分的帧内预测模式选择方法,和上述的H.264/AVC帧图像色度部分的幀内预测模式选择方法采用本发明的H.264/AVC帧图像的帧内预测模式选择的实现方法,由于先行计算宏块失真度确定宏块预测模式的尺寸,初步减少了运行RDO的次数;然后再采用基于SATD的计算进一步获得优选的5 预测模式,最后再在这些优选的预测模式中进行RDO运算,因此有效的减少了运行RDO的次数。 The present invention also provides a method of mode selection H.264 / AVC frame image, which comprises the above-described intra-prediction mode of H.264 / frame image luminance portion AVC 25 selection method, and the above-described H.264 / AVC frame image implementation method of selecting an intra prediction mode using the chrominance portion of H.264 / AVC frame image according to the present invention, the selected intra prediction mode, since the first distortion calculating sized macroblock macroblock prediction mode, the initial operation is reduced RDO times; and then 5-prediction modes further preferably obtained based on SATD calculation, then the last operation in these preferred RDO prediction mode, thus effectively reducing the number of runs of RDO. 由于失真度和SATD的计算都比RDO 的运算要简单得多,因此本发明可以大幅降低运算复杂度,从而明显的减少了编码耗时,而且编码后对视频序列在图像质量和比特率方面的负面影响都可以忽略。 Due to the distortion than the SATD calculation and RDO calculation is much simpler, and thus the present invention can significantly reduce the computational complexity, thereby significantly reducing the time consuming coding, and the coded video sequence in terms of image quality and the bit rate of negative effects can be ignored. 因此,很好的解决了现有技术中H.264/AVC io 视频序列编码过于复杂而无法广泛应用的问题,使得H.264/AVC标准可以很好的应用于实时操作,如网络传输和视频会议等等。 Therefore, a good solution to the problem of the prior art H.264 / AVC io video sequences encoding too complex to be widely used, making the H.264 / AVC standard can be well applied in real-time operations, such as video and network transmission meetings and so on. 附图说明图1为对一段视频序列进行亮度部分的帧内预测模式选择的流程图;15 图2为表示,基于RDO得出的最佳模式是基于SATD选出的n个优选的预测模式之一的概率,其函数关系图;图3为釆用本发明的模式选择方法与釆用标准RDO方法,在编码耗时方面的对比情况柱状图。 BRIEF DESCRIPTION OF DRAWINGS FIG 1 is a flowchart of an intra prediction mode of the luminance of the selected portion of a video sequence; FIG. 2 is a 15, based on a best RDO mode is derived based on the n SATD preferably selected prediction modes a probability function which; Figure 3 is a preclude to preclude the use of standard selection methods RDO method mode of the invention, in comparison of encoding time-consuming aspect of the histogram. 图4为釆用本发明的模式选择方法与釆用标准RDO方法,在平20均图像质量方面的对比情况柱状S 。 4 is to preclude the selection methods preclude the use of standard methods using RDO mode of the invention, the comparison of the columnar S 20 average level of image quality. 图5为釆用本发明的模式选择方法与采用标准RDO方法,在编码比特率方面的对比情况柱状图。 5 is preclude the selection method and the mode of the invention using standard RDO method, in comparison of encoding bit rates of the histogram. 具体实施方式本发明提出的H.264/AVC帧图像发的帧内预测模式选择的方法, 25 结合附图和实施例说明如下。 H.264 intra prediction mode selection method proposed embodiment of the present invention / AVC hair frame image, 25 in conjunction with the drawings and the following examples illustrate. 以下实施方式仅用于说明本发明,而并非对本发明的限制,有关技术领域的普通技术人员,在不脱离本发明的精神和范围的情况下,还可以做出各种变化和变型,因此所有等同的技术方案也属于本发明的范畴,本发明的专利保护范围应由各权利要求限定。 The following embodiments are merely illustrative of the present invention, and are not restrictive of the invention, relating to ordinary skill in the art, without departing from the spirit and scope of the present invention, can make various changes and modifications, all equivalent technical solutions also within the scope of the present invention, the scope of protection of the present invention should be defined in the claims. 5 以下各个实施例的实验条件为:(1) 测试序列总帧数为300;(2) 测试序列的帧率为30帧/秒;(3) 每间隔5帧进行一次编码,因此买际进行帧图像预测编码的帧数为50;io (4)使用帧内预测方式进行编码;(5) 量化参数设为28;(6) 对比的标准编码测试,使用JVT提供的JM98编解码模块(即JVT Reference Software uno伍cial version JM9.8,软件下载地址为http:〃bs.hhi.de/〜suehring /tml/download ); 15 ( 7 )测试釆用的视频序列为JVT测试模型专题小组在2003年2月发布的《ISO/IEC JTC1/SC29/WG11与ITU-T SGI6 Q.6草案第4版》 中,《运动估计的评估》中推荐的典型序列;(8 )实验釆用的计算机配置为P4 Celeron 2.66GHz的CPU, 256M 的内存。 5 each of the following experimental conditions as Example embodiments: (1) the total number of frames of the test sequence 300; frame rate (2) a test sequence of 30 frames / sec; (3) once every 5 frames encoding, thus actually performed buy frame image predictive coding to frames 50; io (4) using the intra-prediction mode encoding; (5) the quantization parameter is set to 28; (6) Comparative test standard code, JM98 using JVT codec module provided (i.e. Wu JVT Reference software uno cial version JM9.8, software download address is http: 〃bs.hhi.de / ~suehring / tml / download); 15 (7) video test sequence preclude the use of focus groups to test model JVT in 2003 in February release of "ISO / IEC JTC1 / SC29 / WG11 and the draft ITU-T SGI6 Q.6 4th Edition", "evaluate motion estimation" recommended typical sequence; computer (8) preclude the use of the experimental configuration as the P4 Celeron 2.66GHz CPU, 256M of memory. 20 以下实施例中进行的测试,用于对比两种方式的结果:其中,一种是采用JM98原来设置对全部宏块模式进行标准编码,另一种采用本文提出的帧图像模式选择的实现方法进行编码。 Tests conducted in Example 20 the following examples, the results for the comparison of two ways: wherein A is provided for all the original JM98 using macroblock mode coding standard, another mode of using the implementation method proposed frame image selected encoded. (实施例一)参照说明书附图1,本实施例用于说明釆用本发明的帧图像亮度25 部分的帧内预测模式选择方法,对一段视频序列进行亮度部分的帧内预测模式选择的流程。 (First embodiment) with reference to the accompanying drawings 1, the present embodiment for explaining a method of selecting preclude the intra prediction mode image brightness portion of the frame 25 of the present invention, a video sequence of intra-prediction mode of the luminance portion of the selection process . 11在本实施例中,首先,把帧图像划分成为相应数目的16x 16大小的宏块,并设定宏块差异度的门限值。 11 in the present embodiment, first, the frame image is divided into a number of corresponding size of 16x 16 macroblock, and sets a threshold value of the degree of difference macroblock. 然后,对每个宏块进行差异度D的计算,在本实施例中,使用15 15 15 15的计算公式是D:ZS〜-(EZ(、V256)2,选择这个公式的好处5 是可以避免开方的运算。关于门限值的确定,在实现本发明的方法的初始阶段,进行大量的测试,用于确定使用不同的门限值进行判决,不同的门限值对应不同的效果,这些效果包括编码时间、编码后的比特率、编码后的图像质量值。综合考虑这些方面之后,确定门限值。在大多数情况下,可10 以把这个值作为一个固定值,在往后的使用中,如果用户没有特殊的要求,可以直接釆用这个值,而不需要重新测试。这个门限值也是可调的,对于特定的情况,例如用户对图像质量的要求特别高,或者传输速率的要求特别高,都可以现有的门限值的基础上对它进行调整, 以适应特定的情况。15 然后进行判决。对于差异度大于门限值的宏块 Then, for each macroblock of the difference D is calculated, in the present embodiment, using the calculation formula is 15 15 15 15 D: ZS~- (EZ (, V256) 2, choose the benefits of this formula is 5 avoid prescribing operation. on the determination threshold value, in the initial stage of implementation of the present invention, extensive testing to determine the use of different threshold values ​​for decision, different threshold values ​​corresponding to different effects, these effects include after encoding time, the bit rate after encoding, the encoded image quality value. considering these aspects, the determination threshold value. in most cases, this value may be 10 as a fixed value, in the next in use, if the user does not have special requirements can preclude the use of this value directly, without the need to re-test. this threshold value is adjustable for a particular situation, such as the user requires a particularly high image quality, or the transmission requires particularly high rate, it can be an existing threshold on the basis of adjustment, to adapt a particular situation and then the decision .15 for difference greater than the threshold macro block ,使用较小的尺寸即4x4的大小进行亮度部分的预测模式选择;对于差异度小于门限值的宏块,使用较大的尺寸即16x16的大小进行亮度部分的预测模式选择。这样的初步选择之后,RDO it行的次数第一次得到减少。在9 种4x4亮度部分的模式中,计算每个模式的SATD值,取其中SATD 20 最小的3种模式,作为优选的预测模式。在4种16x16亮度部分的模式中,计算每个模式的SATD值,取其中SATD值最小的2种模式, 作为优选的预测模式。经过这样计算得出的亮度部分优选的预测模式一共是2种或3种。色度部分方面,由于在H.264AWC中仅支持4种8x8色度部分25的预测模式,因此没有必要利用每一个宏块的差异度来进行预测模式尺寸大小的选择,相对亮度部分的预测模式选择,色度部分得要简单一些。根据色度采样值可以得到目标宏块各个预测模式的SATD值, 在本实施例中,取其 , I.e., the size of the smaller sized 4x4 prediction mode of the luminance portion of the selection; for differences less than the threshold value of the macroblock, i.e., using a larger size of the size 16x16 prediction modes of the luminance portion of the selection of such preliminary selection Thereafter, the number of the first row RDO it is reduced in nine modes 4x4 luminance portion, the SATD value is calculated for each mode, which takes a minimum SATD 20 is three modes as a preferred mode of prediction. of the four 16x16 luminance pattern portion, the SATD value is calculated for each mode, the minimum SATD value which takes two modes as a preferred mode of prediction. after the calculated luminance portion such preferred prediction mode of a total of two or three kinds of . chrominance portion, since the H.264AWC only supports four 8x8 chroma prediction mode portion 25, there is no need to use difference of each macroblock of a prediction mode to select the size of the predicted relative luminance portion mode is selected, the chrominance portion is simpler to obtain. SATD value can be obtained respective target macroblock prediction mode according chroma samples, in the present embodiment, whichever SATD值最小的2种预测模式,所以优选的预测模式是2种。然后对这些优选的预测模式进形RDO运算,获得最优模式。这样,原本需要进行的多次RDO,在本实施例中只需要2-3次。因此, 大大减少了运算量,节约编码耗时。表1给出了标准编码和采用本发明的编码,运行RDO的次数。 可以看出,釆用本发明的方法进行编码,运行RDO的次数明显减少。<table>table see original document page 13</column></row> <table>10 表1(实施例二)本实施例用于说明釆用现有技术所获得的最佳预测模式,正好是釆用基于SATD的计算所获得的几个优选的预测模式中的一个的概率。 SATD value and the smallest prediction modes, it is preferable that two kinds of prediction modes are then predicted for these preferred mode RDO calculation into shape to obtain the optimal mode. Thus, the need for multiple RDO otherwise, in the present embodiment requires only 2-3 times. Thus, greatly reducing the amount of computation saving time consuming coding. table 1 gives the standard encoding and the present invention, the number of runs of RDO can be seen, preclude performed using the method of the present invention encoding, significantly reduced number of runs RDO. <table> table see original document page 13 </ column> </ row> <table> 10 in table 1 (Example II) for explaining the present embodiment preclude the use of the prior art obtained the best prediction mode, with just a few calculated based on the probability of the obtained SATD preferred mode of a prediction of Bian. 具体的实验结果参看图2,图2的横坐标表示基于SATD选出的15 优选的预测模式的个数(n),纵坐标表示的是基于RDO得出的最佳预测模式刚好落在这n个模式中的概率。 Specific results Referring to Figure 2, the abscissa represents the basis of the number of FIG. 2 (n) 15 is preferably selected from the SATD prediction mode, and the ordinate represents the optimum prediction mode based RDO stars just falls on the n a mode probability. 本实施例采用了4个典型序列包括"foreman"、 "clarie"、 "news"、 "carphone"。 The present embodiment employs four exemplary sequences include "foreman", "clarie", "news", "carphone". 其中,横轴表示使用SATD计算来选择的优选的预测模式的个数,纵轴表示当前的模式个数的情况下,采用现有技术所获得20 的最佳预测模式,正好是采用基于SATD的计算所获得的几个优选的预测模式中的一个的概率。 Wherein, the horizontal axis represents the number of preferred use SATD calculating prediction mode is selected, and the vertical axis represents the case number of the current mode, using the prior art optimum prediction mode 20 is obtained, based exactly the SATD calculating a probability obtained by several preferred prediction mode. 从图中可以看出,对于不同的4个测试序歹"J,采用基于SATD的计算来选择优选的预测模式时,当选择的模式个数大于或等于3时,这些优选的预测模式与采用现有技术所得的最佳模式,90%以上的概率是等同的,而其中,对于比较好的情况,例如序列"news", 97°/。的概率是等同的。也就是说,对于本发明的方法, 5 釆用基于SATD的计算获得优选的预测模式,当所选择的预测模式个数不小于3时,再对这些模式进行RDO运算获得最佳模式,相比起现有技术的标准方法,两者的效果在非常大程度上是等同的。由于SATD的计算相对于RDcost的计算要简单得多,因此,采用本发明的方法,能在很大程度上减少运算量,节省运算时间。io (实施例三)本实施例用干比较采用不同的差异度门限值,对图像编码的效果影响。因为差异度门限值的确定在本发明的方法中很关键,所 As can be seen from the figure, four different test sequences bad "J, using the SATD calculation based on the selected prediction mode is preferred, when the number of the selected mode is greater than or equal to 3, such preferred prediction mode using the resulting prior art the best mode, the probability that more than 90% identical, and wherein for a better situation, such as a sequence "news", 97 ° /. is equivalent to the probability. That is, for the present invention the method, based on the calculated SATD 5 preclude the use of the prediction mode to obtain the preferred, when the number of selected prediction mode is not less than 3, then those patterns best mode RDO calculation, compared to the standard methods of the prior art, effects of the two in a very large extent are identical. Since the calculated relative to calculate SATD RDcost much simpler, therefore, the method of the present invention, the amount of calculation can be reduced to a large extent, saving computation time .io (Third embodiment) the present embodiment was used to compare different dry difference degree threshold, influence on the image encoding results because of the difference in the threshold determination is critical in the process of the present invention, the 本发明的初始阶段,进行了大量的测试,获得大量的测试数据,针对不同的门限值所获得的不同的编码效果。限于篇幅,本实施例仅列出了5 15 个不同的门限值Do,以此来说明不同门限值对最终编码性能的影响, 具体的评价参数包括编码耗时的差异、图像质量的影响、比特率变化。 其中,编码耗时变化的负号表示釆用本发明的方法编码后,相对于标准编码所节省的时间百分比;图像质量的负号表示釆用本发明的方法编码后,相对于标准编码在图像质量方面的下降分贝。20 表2给出了本实施例的结果,测试序列为"Foreman"。可以看出,门限值Do设置得越大,编码时间就越少,编码所需的时间越短, 实时性就会越好,但图像质量会变差。因此用户或系统可以根据图像质量的具体要求和网络的情况,设置具体的门限值,综合考虑当前情况所需得图像质量和编 The initial stage of the present invention, a lot of tests, a large amount of test data obtained, the effect of different codes for different thresholds obtained. Due to space limitations, the present embodiment only lists 515 different threshold values do, in order to illustrate the effect of different threshold values ​​on the final coding performance, specific encoding parameters include the time-consuming evaluation of the differences, affecting the image quality, the bit rate variation. wherein the encoded time-consuming change preclude the use of this negative sign indicates after the encoding method of the invention, the standard code with respect to the time saved percentage; negative image quality represented preclude the use of the method for encoding according to the present invention, with respect to the standard coding dB drop in image quality of .20 table 2 shows the present results embodiment, the test sequence is "Foreman". as can be seen, the larger the threshold value Do is provided, the less encoding time, the shorter the time required for encoding, the better real-time, but the image quality deteriorated. Thus the user or an image system according to the specific requirements and the quality of the network, setting a specific threshold value, considering the current situation to obtain a desired image quality and encoding 码耗时,以达到合适的效果。<table>table see original document page 14</column></row> <table><table>table see original document page 15</column></row> <table>表2从表2可以看出,D(Tl50的情况下,编码耗时可以节省40%以上,而图像质量的下降只有0.04分贝,比特率的变化也只有0.4%, 这样的负面影响是可以忽略的。因此,在往后的使用中,只要用户对5 编码效果没有特殊的要求,那么Dfl50可以作为一个相对固定的门限值,不需要用户再重复测试以获得门限值。 Code consuming, to achieve proper results. <Table> table see original document page 14 </ column> </ row> <table> <table> table see original document page 15 </ column> </ row> <table> table 2 As can be seen from table 2, the case where D (Tl50, coding can save 40% or more time-consuming, while the degradation of image quality is only 0.04 db, bit rate change is only 0.4%, so that negative effects are negligible Therefore, in subsequent use, as long as the effect of encoding the 5 user no special requirements, then Dfl50 relatively fixed as a threshold value, the user does not need to repeat the test to obtain a threshold value. 如果用户对编码效果有特殊的要求,例如需要获得特别清晰的图像质量,则可以在Dfl50 的基础上调整这个门限值,例如使用0。 If a user has special requirements for coding results, such as the need to obtain a particularly clear image quality, this threshold value can be adjusted on the basis Dfl50, for example, 0. =50;如果用户对传输速度的要求特别高,却不要求精确的图像质量,可以使用髙一些的门限值, io 例如D『400。 = 50; if the user requires particularly high transmission speed, but do not require precise image quality, can be used Gao some threshold, e.g. IO D "400. (实施例四)本实施例用于说明针对不同的典型序列,釆用本发明的模式选择方法所获得的结果,相对于现有技术的标准方法,在编码耗时、图像质量损失、比特率变化方面的比较。 (Fourth Embodiment) of the present embodiment for illustrating a typical embodiment of the different sequences, with the result preclude mode selection method of the present invention is obtained, with respect to the standard methods of the prior art, in coding time-consuming, loss of image quality, bit rate comparison of change. 其中所用的6个测试序列分别是15 "News"、 "Foreman"、 "Bike,,、 "Mobile" 、 "Bus"、 "Silent"。本实施例中所用的门限值为D0=150,从实施例3可以看出,这个值是对编码耗时、图像质量和比特率变化的一个折中。在没有特殊要求的情况下, D0=150可以作为一个固定值在大多数情况下使用,因此本实施例中选用了这个值。20 具体的测试数值见表3。其中,耗时变化的负号表示釆用本发明的方法编码后,相对于标准编码所节省的时间百分比;图像质量的负号表示釆用本发明的方法编码后,相对于标准编码在图像质量方面的下降分贝。<table>table see original document page 16</column></row> <table>表3从结果数据可见,在门限值为150的情况下,对于比较好的情况, 5 编码耗时最多可节省68.5% (序列"Silent"),即使是比较不理想的情况,也可以节省30% (序列"News")。而无论对于哪个测试序列, 6 wherein the used test sequences are 15 "News", "Foreman", "Bike ,,," Mobile "," Bus "," Silent ". Threshold used in the examples of the present embodiment is D0 = 150, from Example 3 as can be seen, this value is in the case where no special requirements, D0 = 150 can be used in most cases for the time-consuming encoding, a compromise image quality and bit rate variation as a fixed value, in the present embodiment, the value chosen .20 Table 3. the specific test value where the negative sign represents a time-consuming change preclude the use of the method for encoding according to the present invention, the standard code with respect to the percentage of time saved; negative image quality numerals preclude the use of the method for encoding according to the present invention, with respect to the standard coding drop dB in terms of image quality. <table> table see original document page 16 </ column> </ row> <table> table 3 data from the results can be seen, in the case where the threshold is 150, for a better case, encoding takes savings 5 ​​68.5% (sequence "Silent") at most, even less desirable case, you can save 30% (sequence "News" ). for regardless of which test sequences, 们的图像质量的影响都不大于0.04分贝,这样的程度的影响是可以忽略的,而比特率的变化也不高于0.5%,同样是可以忽略的。可见, 釆用本发明的方法,可以在节省运算量的情况下,明显的减少编码耗io 时,同时几乎不会对图像质量或比特率造成不可忽略的影响。图3显示了对测试序列"Foreman",采用现有技术的RDO标准方法和釆用本发明的方法对测试序列进行编码后,对于不同的编码帧数,在编码耗时方面的比较。其中,有斜线填充的柱体表示现有技术的标准方法的情况,无填充的柱体表示本发明的方法的情况。横轴表15示不同的编码帧数,单位是"帧",纵轴是编码耗时,单位是"秒"。图4显示了对测试序列"Foreman",釆用现有技术的RDO标准方法和采用本发明的方法对测试序列进行编码后,随着对于不同的编码帧数,在图像质量方面的比较。其中,有斜线填 It affects image quality is not more than 0.04 dB influence the extent is negligible, and does not change a bit rate higher than 0.5%, the same is negligible. Visible, preclude the use of the method according to the present invention, in the case of saving the amount of calculation, a significant reduction in the time of encoding IO consumption, while almost no impact on image quality can not be ignored or bit rate. FIG. 3 shows a sequence of test "Foreman", the prior art using standard RDO Bian coding method and after the test sequence by the method of the present invention, coding for the different frames, in comparison of encoding time-consuming aspect. wherein hatched column packed shows the case of the standard prior art methods, no column packed shows the case of the method of the present invention. the horizontal axis shows different coding frame 15, the unit is "frame", and the vertical axis is time consuming coding, unit is "second". FIG. 4 shows the test sequence " Foreman ", RDO standard methods and method of the present invention preclude the prior art test sequence after encoding with different coding frame, in terms of image quality comparison. wherein hatched fill 的柱体表示现有技术的标准方法的情况,无填充的柱体表示本发明的方法的情况,横轴表示不同的编码帧数,单位是"帧,,,纵轴是图像质量,单位是"分贝"。 Column shows the case of the standard prior art method, represented by unfilled cartridge case of the method of the present invention, the horizontal axis represents the different coding frames, the unit is "frame ,,, vertical axis is the image quality, the unit is "decibel". 图5显示了对测试序列"Foreman",采用现有技术的RDO标准方法和采用本发明的方法对测试序列进行编码后,随着对于不同的编码帧数,在比特率方面的比较。 Figure 5 shows the test sequence "Foreman", the RDO standard methods and the present invention is a method of using the prior art test sequence encoded with different coding frame, the bit rates of the comparison. 其中,有斜线填充的柱体表示现有技术5 的标准方法的情况,无填充的柱体表示本发明的方法的情况。 Wherein hatched column packed shows the case of the prior art standard methods. 5, unfilled column shows the case of the method of the present invention. 横轴表示不同的编码帧数,单位是"帧",纵轴是比特率,单位是"Kbps"。 The horizontal axis represents the different coding frames, the unit is "frame", and the vertical axis is the bit rate, the unit is "Kbps." 可以看出,采用本发明的方法和采用现有技术的标准RDO的方法对测试序列进行编码,两者的结果在图像质量和比特率方面基本相同,但是编码耗时方面,釆用本发明的方法显然能大幅节省时间,实io 际所用耗时只需标准RD O的方法的41 %到4 6 %的编码时间。 As can be seen, the method of the present invention and the method of using RDO standard prior art test sequence encoding results of both image quality and bit rates is substantially the same, but time-consuming aspect of the encoding, preclude the use of the present invention. the method can obviously significant savings in time, only the real io international standard RD O Processed method used 41% to 46% of the encoding time.

Claims (9)

1. 一种H.264/AVC帧图像亮度部分的帧内预测模式选择方法,其特征在于,它包括以下步骤: (1)将视频帧图像划分为相应数目的16×16大小的宏块; (2)设定视频帧图像宏块的差异度门限值; (3)计算视频帧图像每个宏块的差异度; (4)比较步骤(3)所得的差异度和步骤(2)所得的门限值; (5)根据不同的比较结果选择不同尺寸的帧内预测模式,并计算每个预测模式所对应的绝对变换差之和SATD的值; (6)根据步骤(5)所得的SATD值,获得优选的预测模式; (7)对步骤(6)所得的优选的预测模式,通过率失真优化,确定最佳模式。 An H.264 / AVC intra prediction mode of a luminance frame image selection portion, characterized in that it comprises the following steps: (1) the image into a corresponding video frame number of the size of 16 × 16 macroblocks; (2) setting the degree of difference video gate frame image macroblock value; (3) calculating the difference for each macroblock of a video frame image; resulting (4) comparing the difference of step and step (2) (3) of the resulting threshold; (5) selecting an intra prediction mode different sizes depending on the comparison result, and the calculated value and the absolute transformed difference between the SATD corresponding to each prediction mode; (6) according to step (5) of the resulting SATD value, preferably obtained prediction mode; (7) of step (6) is preferably obtained prediction mode by rate distortion optimization, to determine the best mode.
2、 如权利要求1所述的帧图像亮度部分的帧内预测模式选择方法,其特征在于,对于步骤(2)的门限值,不是一个固定值,根据15不同的编解码图像质量和传输速率的要求来调整门限值的具体值。 2. The method claimed in selecting an intra prediction mode of the luminance portion of the image frame of claim 1, wherein, for step (2) of the threshold not a fixed value, 15 different codec and transmission in accordance with the image quality DETAILED rate required to adjust the value of the threshold value.
3、 如权利要求2所述的帧图像亮度部分的帧内预测模式选择方法,其特征在于,步骤(3 )中差异度的计算使用公式Z^cre戸wq^ 22"砂-(12^;)/256)2 其中Z表示的是一个宏块中的象素点的亮度值;如果用一个宏块中的象素点的亮度值构造一个矩20 阵,则差异度Discrepancy就是该矩阵的标准方差。 3, the intra prediction mode as claimed in the frame image luminance portion selection method of claim 2, wherein the calculating step (3) of the difference using the formula Z ^ cre Kobe wq ^ 22 "sand - (12 ^; ) / 256) 2 wherein Z represents the luminance value of pixel dots of a macro block; if the luminance value of pixel dots of a configuration of a macroblock moment matrix 20, the degree of difference of the matrix is ​​the standard discrepancy variance.
4、如权利要求2所述的帧图像亮度部分的帧内预测模式选择方法,其特征在于,步骤(3)中差异度的计算公式是"-2^(、-(ZZ(^)/256)2其中丄表示的是一个宏块中的象素:t=0 "0 x=0 "0 矽点的亮度值;这个差异度的值也表示,如果用宏块中的象素点的亮度值构造一个矩阵,则差异度D就是该矩阵的方差。 4. The method claimed in selecting an intra prediction mode of the luminance portion of the image frame of claim 2, wherein the step (3) the calculation formula of the difference is "-2 ^ (, - (ZZ (^) / 256 ) 2 wherein Shang is represented by pixels in a macroblock: t = 0 "0 x = 0" point of silicon luminance value 0; the value of the said difference, if the pixel dots macroblock luminance Constructs a matrix, D is the variance of the difference of the matrix.
5、 如权利要求3或4所述的帧图像亮度部分的帧内预测模式选择方法,其特征在于,步骤(5)所选择的不同尺寸的帧内预测模式, 5有16xl6的帧内预测模式,和4x4的帧内预测模式;对于差异度小于门限值的宏块,选用16x16的帧内预测模式,对于差异度大于门限值的宏块,选用4x4的帧内预测模式;对于这两种尺寸的预测模式,计算每一种预测模式对应的SATD值。 5. The method claimed in intra prediction mode selection frame image luminance portion 3 or claim 4, wherein the step (5) different sizes of the selected intra prediction mode, intra prediction mode 5 has a 16xl6 , and 4x4 intra prediction mode; macroblock for differences less than the threshold value, the selection of an intra 16x16 prediction mode for the macroblock difference is greater than the threshold, the 4x4 intra prediction mode selection; for two sizes prediction mode calculates SATD value corresponding to each prediction mode.
6、 如权利要求1所述的帧图像亮度部分的帧内预测模式选择方io 法,其特征在于,根据步骤(5)所得的SATD值,对9种4x4亮度预测模式,取其中SATD值最小的3种模式,作为优选的预测模式; 对4种16x16亮度预测模式,取其中SATD值最小的2种模式,作为优选的预测模式。 6, as claimed in the intra prediction mode of the luminance portion of the image frame of claim 1 io method selector, wherein, according to SATD value obtained in step (5), for nine 4x4 luma prediction mode, which takes a minimum SATD value the three modes as a preferred mode of prediction; on four 16x16 luma prediction mode, the minimum SATD value which takes two modes as a preferred mode of prediction.
7、 如权利要求1所述的帧图像亮度部分的帧内预测模式选择方15法,其特征在于,步骤(1)和步骤(2)的操作顺序可以互换。 7, as claimed in the intra prediction mode of the luminance portion of the image frame of claim 1 method selector 15, characterized in that the operating order of the steps (1) and step (2) may be interchanged.
8、 一种11264/八¥(:帧图像色度部分的帧内预测模式选择方法, 其特征在于,它包括以下步骤:(1) 将视频帧图像划分为相应数目的16xl6大小的宏块;(2) 计算视频帧图像每个宏块的SATD值; (3)根据步骤(2)所得的SATD值,对4种8x8色度帧内预测模式,取其中SATD值最小的2种模式,作为优选的预测模式;(4)对步骤(3)所得的预测模式,通过率失真优化,确定最佳模式。 8 A 11264 / ¥ eight (: intra-prediction mode of the chrominance portion selected frame image, characterized in that it comprises the following steps: (1) the image into a video frame sizes 16xl6 a corresponding number of macroblocks; (2) a video frame image SATD value is calculated for each macroblock; (3) according to step (2) obtained in SATD value, chroma on four 8x8 intra prediction modes, of which the lowest of SATD value 2 modes, as the preferred mode of prediction; (4) step (3) prediction mode obtained through rate distortion optimization, to determine the best mode.
9、 一种11.264/八¥(:帧图像的模式选择方法,其特征在于,它包25 括如权利要求1所述的H.264/AVC帧图像亮度部分的帧内预测模式选择方法,和如权利要求8所述的H.264/AVC帧图像色度部分的帧内预测模式选择方法。 9. A 11.264 / eight ¥ (: mode selection method of the frame image, characterized in that it comprises a package 25 as H.264 intra prediction mode selection method of claim 1 / frame image luminance portion as claimed in claim AVC, and H.264 / AVC intra prediction mode of the chrominance portion of the frame image as claimed in claim 8 selected method.
CN 200510086742 2005-10-31 2005-10-31 Method of selecting in frame prediction mode based on H.264/AVC standard frame image CN100417228C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200510086742 CN100417228C (en) 2005-10-31 2005-10-31 Method of selecting in frame prediction mode based on H.264/AVC standard frame image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200510086742 CN100417228C (en) 2005-10-31 2005-10-31 Method of selecting in frame prediction mode based on H.264/AVC standard frame image

Publications (2)

Publication Number Publication Date
CN1753500A CN1753500A (en) 2006-03-29
CN100417228C true CN100417228C (en) 2008-09-03

Family

ID=36680149

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200510086742 CN100417228C (en) 2005-10-31 2005-10-31 Method of selecting in frame prediction mode based on H.264/AVC standard frame image

Country Status (1)

Country Link
CN (1) CN100417228C (en)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101888549A (en) * 2010-06-18 2010-11-17 浙江大学 Intra-frame 4*4 forecasting model selection method based on transform domain information

Families Citing this family (15)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101087427B (en) 2006-06-06 2011-04-06 北京大学深圳研究生院 A H.264 standard in-frame prediction mode selection method
US8218636B2 (en) 2006-11-21 2012-07-10 Vixs Systems, Inc. Motion refinement engine with a plurality of cost calculation methods for use in video encoding and methods for use therewith
CN101198062A (en) * 2006-12-04 2008-06-11 华为技术有限公司;清华大学 Method and system for estimating compression encoding
CN100571390C (en) 2006-12-21 2009-12-16 联想(北京)有限公司 Method and apparatus for fast mode selecting of H264 video coding
CN100479523C (en) 2007-05-16 2009-04-15 华为技术有限公司 Detecting alarm method and alarm system
CN100566427C (en) * 2007-07-31 2009-12-02 北京大学;北京北大方正电子有限公司 Intraframe prediction coding optimum mode selecting method and apparatus for video coding
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
CN100579227C (en) 2007-12-25 2010-01-06 上海广电(集团)有限公司中央研究院 System and method for selecting frame inner estimation mode
US9055305B2 (en) 2011-01-09 2015-06-09 Mediatek Inc. Apparatus and method of sample adaptive offset for video coding
CN106454357A (en) * 2011-01-09 2017-02-22 寰发股份有限公司 Apparatus and method of sample adaptive compensation of processed video data
US9161041B2 (en) 2011-01-09 2015-10-13 Mediatek Inc. Apparatus and method of efficient sample adaptive offset
JP2013034162A (en) * 2011-06-03 2013-02-14 Sony Corp Image processing device and image processing method
CN103327325B (en) * 2013-05-13 2016-05-25 西安电子科技大学 The quick self-adapted system of selection of intra prediction mode based on HEVC standard
CN103442235B (en) * 2013-09-06 2017-01-25 融创天下(上海)科技发展有限公司 Image processing method and device
CN106231307B (en) * 2016-08-09 2019-05-03 赵羽剑 A kind of compression of images intra-coding prediction method and its hardware realization

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1556652A (en) 2003-12-30 2004-12-22 中国科学院计算技术研究所 Video freguencying frame image fast coding method based on optimal prediction mode probability
CN1585495A (en) 2004-06-11 2005-02-23 上海大学 Quick selection of prediction modes in H.264/AVC frame

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1556652A (en) 2003-12-30 2004-12-22 中国科学院计算技术研究所 Video freguencying frame image fast coding method based on optimal prediction mode probability
CN1585495A (en) 2004-06-11 2005-02-23 上海大学 Quick selection of prediction modes in H.264/AVC frame

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
H.264/AVC 帧内预测模式选择算法研究. 裴世保,李厚强,俞能海.计算机应用,第25卷第8期. 2005

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101888549A (en) * 2010-06-18 2010-11-17 浙江大学 Intra-frame 4*4 forecasting model selection method based on transform domain information
CN101888549B (en) 2010-06-18 2012-05-02 浙江大学 Intra-frame 4*4 forecasting model selection method based on transform domain information

Also Published As

Publication number Publication date
CN1753500A (en) 2006-03-29

Similar Documents

Publication Publication Date Title
CN1332563C (en) Coding method of video frequency image jump over macro block
Huang et al. Analysis and reduction of reference frames for motion estimation in MPEG-4 AVC/JVT/H. 264
KR100984517B1 (en) Fast mode decision making for interframe encoding
CN101175209B (en) Adaptive de-blocking filtering method for mpeg video decoder
CN100399829C (en) Video transcoding method and apparatus and motion vector interpolation method
CN1206864C (en) Association rate distortion optimized code rate control method and apparatus thereof
CN101171843B (en) Content classification for multimedia processing
CN100566426C (en) Method and apparatus for encoding and decoding of video
KR20110113592A (en) Method and apparatus of performing intra prediction using adaptive filter
JP5563042B2 (en) Adaptive video frame interpolation
CN101138250B (en) Adaptive intra-frame refresh for digital video encoding
CN101313587B (en) Mode selection techniques for multimedia coding
KR100942395B1 (en) Rate control for multi-layer video design
US9961372B2 (en) Adaptive disabling of deblock filtering based on a content characteristic of video information
CN101232618B (en) Method and device for indicating quantizer parameters in a video coding system
CN101189882B (en) Method and apparatus for encoder assisted-frame rate up conversion (EA-FRUC) for video compression
CN100401789C (en) Quick selection of prediction modes in H.264/AVC frame
JP2006519564A (en) Video encoding
CN101406056B (en) Method of reducing computations in intra-prediction and mode decision processes in a digital video encoder
CN1225126C (en) Space predicting method and apparatus for video encoding
EP2202986A1 (en) An adaptive predicting device and method
CN101015215B (en) Methods and apparatus for performing fast mode decisions in video codecs.
TWI230547B (en) Low-complexity spatial downscaling video transcoder and method thereof
US20060165163A1 (en) Video encoding
KR101299421B1 (en) System and method of coding mode decision for video encoding

Legal Events

Date Code Title Description
PB01 Publication
C06 Publication
SE01 Entry into force of request for substantive examination
C10 Entry into substantive examination
GR01 Patent grant
C14 Grant of patent or utility model
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20080903

Termination date: 20181031

CF01 Termination of patent right due to non-payment of annual fee