CN100409690C - Video interframe compression method based on space-time correlation - Google Patents

Video interframe compression method based on space-time correlation Download PDF

Info

Publication number
CN100409690C
CN100409690C CN 200610080881 CN200610080881A CN100409690C CN 100409690 C CN100409690 C CN 100409690C CN 200610080881 CN200610080881 CN 200610080881 CN 200610080881 A CN200610080881 A CN 200610080881A CN 100409690 C CN100409690 C CN 100409690C
Authority
CN
China
Prior art keywords
macroblock
correlation
coding
video
method
Prior art date
Application number
CN 200610080881
Other languages
Chinese (zh)
Other versions
CN1845608A (en
Inventor
刘鹏宇
晶 谢
贾克斌
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 200610080881 priority Critical patent/CN100409690C/en
Publication of CN1845608A publication Critical patent/CN1845608A/en
Application granted granted Critical
Publication of CN100409690C publication Critical patent/CN100409690C/en

Links

Abstract

基于时空相关性的视频帧间压缩方法,涉及视频压缩领域。 Interframe video compression based on temporal correlation method, relates to video compression. 本发明的特征在于具有以下处理步骤:(1)取出原始的视频数据,针对其中的亮度分量处理;(2)得到一帧图像中的一个宏块的亮度值;(3)计算宏块的空间相关性,利用水平、垂直预测方法得到当前宏块与预测宏块的残差值,作为空间相关性的度量;(4)计算宏块的时间相关性,得到当前宏块与前一帧对应位置宏块的差值,作为时间相关性的度量;(5)比较空间相关性和时间相关性的大小;(6)根据比较后的结果,采用相应的编码模式对该宏块编码。 Feature of the present invention is characterized by having the following processing steps: (1) remove the original video data, the luminance in which the component to be processed; (2) to obtain the luminance value of an image of a macro block; space (3) calculation of the macroblock correlation using horizontal, vertical prediction method and the prediction residue values ​​to obtain the macroblock, as a measure of the spatial correlation of the current macroblock; time (4) calculated correlation macroblock to obtain the current position of the previous frame and the corresponding macro block a difference macroblock, as a measure of the time correlation; size (5) comparing spatial correlation and a temporal correlation; (6) based on a result of comparison, the corresponding coding mode for the macroblock coding. 本发明所提供的方法与视频编码标准H.264中采用的帧间编码方法相比,在压缩比相差很少的情况下,能够大幅度提高帧间编码的速度。 The method of the present invention provides inter-coding method compared to the H.264 video coding standard employed in the case of a small difference between the compression ratio, it can greatly improve the speed of inter-coding.

Description

基于时空相关性的视频帧间压縮方法技术领域本发明涉及视频压缩领域,设计和实现了一种基于时空相关性的视频压縮方法。 The present invention relates to the field of video compression based on temporal correlation between frames of video compression FIELD, design and implement a video compression method based on spatio-temporal correlation. 背景技术随着近年来多媒体技术的广泛应用,作为多媒体信息技术的基础,视频压縮方法发挥着重要和不可替代的作用,它可以在很大程度上减轻对存储媒体容量和传输带宽的要求。 Background Art With the wide application of multimedia technology in recent years, as a basis for multimedia information technology, video compression methods play an important and irreplaceable role, which can reduce the requirements for transmission bandwidth and storage capacity of the media to a large extent. 信息技术对视频压縮方法提出的要求是压縮比越来越高、压縮质量越来越好、压縮速度越来越快。 IT requirements for video compression method proposed is getting higher and higher compression ratio, compression quality is getting better, faster and faster compression. 由于视频信息之间存在冗余信息,利用视频信息自身所具有的相关性, 通过某种特殊的技术,能够去除视频数据中的冗余信息,达到视频压縮的目的。 Because of the redundancy of information between the video information, using the video information having the relevance itself, through a special technique, can remove redundant information in the video data reach the video compression purposes. 在视频压縮技术中主要采用I帧(帧内)和P帧(帧间)编码技术,I帧是视频中的某一帧单独编码,与其它相邻帧没有关系,而P帧则是利用前后帧的相关性来编码的,通过这两种方法去除冗余,达到压縮的目的。 In video compression techniques employed in the main I-frame (Intra) and P-frames (inter-frame) coding technique, a single I frame is a frame of video coding, there is no relationship with the other adjacent frame, while the P-frame is to use correlation before and after the encoded frame, removing redundant by the two methods, to achieve the purpose of compression. 由于标准化的需要,近十年中,出现了很多的视频编码标准。 Because of the need for standardization, in recent years, there have been many video coding standards. H.264视频编码标准是由ISO和ITU建立的联合视频工作组(Joint Video Team,简称JVT)制定的,它在ISO中的正式名称是MPEG-4标准的第十部分,在ITU中的正式名称是H.264。 H.264 video coding standard is the Joint Video Team established by ISO and ITU (Joint Video Team, referred to as the JVT) to develop its official name is the ISO standard MPEG-4 Part X, an official of the ITU name is H.264. H.264最大的特点是具有极高的压縮比, 良好的视频质量,但是压縮比的提高是以增加算法复杂度为代价的,致使其编码速度在通常情况下还不能满足实际应用的要求。 H.264 biggest characterized by a high compression ratio, good video quality, but increase the compression ratio is increased at the expense of the complexity of the algorithm, resulting in its encoding speed under normal circumstances can not meet the actual application Claim. 因此,在不改变H.264 标准码流结构和维持原有码率的情况下,找到一种新的方法来替代原有的复杂算法成为研究热点。 Thus, without changing the H.264 standard stream structure and maintain the original bit rate, to find a new way to replace the original complex algorithms become a hot topic. 在H.264引入的众多新技术中,帧间预测编码技术与其它编码标准所釆用的技术有很大的不同。 In many new technologies introduced in H.264 inter prediction coding techniques to other coding standards preclude the use of very different technologies. 在H.264中,对需要编码的宏块采用两种不同的模式来编码, 一种是传统的运动估计与补偿方法,另一种是帧内预测方法。 In H.264, the macroblock to be encoded using two different encoding modes, one is the traditional method of motion estimation and compensation, the other is the intra prediction method. 对于标准中采用的帧内预测方法来说,使用Intra4X4模式和Intral6X 16模式,这两种模式是帧内空间预测的方法,利用的是相邻像素所具有的相关性。 For intra prediction method employed in the standard, using Intra4X4 mode and Intral6X 16 mode, which two modes of the method of intra spatial prediction using the neighboring pixels has a correlation. 对于帧间编码来说,为了减少运动估计后所得到的残差值,R264提供了4种宏块分割模式,划分大小分别为:16X16, 16X8, 8X16和8X8,其中8X8 模式又称为亚宏块模式,它还可以继续进行划分为8X8, 8X4, 4X8和4X4, 如图1所示。 For inter-coding, in order to reduce the residual value obtained after motion estimation, R264 provides four macroblock partition mode, are divided into size: 16X16, 16X8, 8X16, and 8X8, 8X8 mode which is also known as sub-macro block mode, it can continue to be divided into 8X8, 8X4, 4X8 and 4X4, as shown in FIG. H.264中的运动搜索是以子块为单位进行的,所以16X16宏块中的每个子块都有一个独立的运动矢量。 The H.264 motion search is performed in units of sub-blocks, each sub-block in the 16X16 macro block has an independent motion vector. 显然,宏块中的子块划分得愈细, 对残差值进行编码所需的位数就愈小,但因此而带来的问题是有更多的运动矢量需要进行编码,这会导致整个宏块编码所需的位数增加。 Obviously, the sub-blocks in the macroblock survived fine division, the number of bits required for encoding the residual value of the less, but the problem is thus brought more motion vectors need to be encoded, which will cause the entire increasing the number of bits required for encoding the macroblock. 因此,在具体编码时需要根据运动估计后的残差值大小和需要的运动矢量编码代价来优化选择宏块的编码模式。 Thus, in accordance with the specific encoding the desired size of the residual values ​​and the motion vector encoding cost required for motion estimation to optimize the selection of the coding mode for the macroblock. 在H.264中采用的是率失真模型,即RD模型来选择最优的模式。 H.264 is used in rate-distortion models, namely RD model to choose the best model. 在H.264的标准算法中,为了得到最佳的编码模式,对一个宏块要同时用两种模式来预测,根据最后的实际效果来得到最佳的编码方法。 In the H.264 standard algorithm in order to obtain optimal encoding mode for a macro block to be predicted in two modes simultaneously, according to the final results to the actual optimal encoding method. 这样做固然可以得到最佳的编码方式,但代价是计算量显著增加,编码速度变得很慢,这就是影响汪264编码速度的原因所在。 Doing so can certainly get the best encoding, but at the cost of a significant increase in computing, coding speed becomes very slow, and this is the reason Wang affect 264 coded speed lies. 发明内容本发明的目的是结合H.264中采用的帧间编码技术,提出一种基于宏块时空相关性的帧间编码方法。 SUMMARY OF THE INVENTION The present invention is a combination of inter-frame encoding technique employed in H.264, provide a spatio-temporal correlation macro-block based inter-coded. 它利用宏块自身所具有的特征来判断宏块的编码模式。 It is also characterized in using its own macroblock coding mode of the macroblock is determined. 对于某个具体的宏块来说,对它的压縮就是利用它周围的宏块及前面帧中宏块之间的相关性来去除宏块间的信息冗余,从而达到用少量的比特表示大量像素信息的目的。 For a particular macroblock, the compression it is using the correlation between macroblocks around it and front frame macroblock remove redundant information among the macroblocks, so as to achieve a small number of bits represented the purpose of a large number of pixel information. 在本发明中,充分利用宏块所具有的这种特性, 实现了一种基于空间和时间相关性大小来判断宏块编码模式的方法。 In the present invention, the full use of such characteristics possessed by macroblock, implements a method based on spatial and temporal correlation to determine the size of the macroblock coding modes. 本发明首先利用简单的预测方法,得到空间预测宏块及时间预测宏块,将两个预测宏块与原始宏块相减,得到两个预测残差值。 First, the present invention is a simple prediction method, to obtain spatial and temporal prediction macroblock prediction macroblock, the prediction macroblock and two of the original macroblocks are subtracted to obtain two prediction residuals. 同时,考虑到除了残差值外, 其他额外信息对宏块编码模式选择的影响,给每个残差值乘以相应的系数, 比较乘系数后的两个值,如果空间相关性大于时间相关性就采用帧内预测的模式,如果时间相关性大于空间相关性就采用运动估计与补偿的模式。 Meanwhile, in addition to taking into account the residual value, the influence of other additional information of the macroblock coding mode selection, for each residual value multiplied by the corresponding coefficient, the comparison value multiplied by two coefficients, if the spatial correlation is greater than the time-related on the use of the intra prediction mode, if the time correlation is greater than the spatial correlation on the use of motion estimation and compensation mode. 这样, 不仅可以充分利用宏块自身的特点,而且利用比较系数可以减少其它额外信息对判断方法的影响。 Thus, not only can take full advantage of the characteristics of the macro block itself, and using the comparison factor to reduce the impact of other additional information determination method. 本发明的技术思路特征为:1. 利用当前宏块与周围相邻宏块的像素之间有相关性的特点,通过垂直、水平预测(见图2,图3),得到一个衡量空间相关性的预测宏块,二者相减进而得到预测后的残差值。 Technical idea of ​​the present invention is characterized in: using a correlation characteristic between the pixels around the current macroblock and neighboring macroblocks by a vertical, horizontal prediction (see FIGS. 2, FIG. 3), to give a measure of spatial correlation the predicted macroblock, and further subtracting the two to obtain the prediction residual values. 2. 利用当前宏块与前一帧中相同位置的宏块在时间上的相关性,以前一帧相应位置上的宏块作为预测宏块,二者相减进而得到预测后的残差值。 2. Using the same position as the current macroblock in the previous frame macroblock correlation in time, the previous macroblocks as a predictive macroblock corresponding position, and further subtracting the two to obtain the prediction residual values. 3. 如果直接比较两个残差值,没有考虑到诸如编码模式信息,运动矢量信息等额外信息对判断的影响,所以在本发明中,分别给两个残差值乘一个系数, 通过这个系数来调节其它信息对比较的影响。 3. If the direct comparison of the two residual value, without taking into account additional information such as coding mode information, motion vector information and the like impact on the determination, so in the present invention, respectively, to the residual value multiplied by a factor two by this factor adjusting the influence of other information comparison. 4. 根据以上得到的修正后的残差值,比较二者的大小,残差值小的代表相关性大,就采用相应的编码模式。 Residual value after the correction obtained according to the above, both the size of the comparison, a small residual values ​​representative of the correlation is greater, with the corresponding coding mode. 5. 在本发明中,充分考虑到所有可能出现的情况,对于特征明显的宏块采用上述方法能够快速且准确的判断出编码模式,但对于特征不明显的宏块如果依然采用上述方法,则很容易产生误判,所以在新方法中仍然采用原来的方法判断。 5. In the present invention, due account of all possible situations, for obvious characteristics macroblock using the above method to quickly and accurately determine the coding mode, but still using the method described above, if the macroblock is not obvious, then it is prone to false positives, so the method is still employed in new method of determining the original. 6. 综合考虑,本发明即可以对特征明显的宏块直接采用相应的方法来判断, 提前得到最佳的编码模式,同时对特征不明显的宏块采用原来的方法,这样充分考虑到所有的情况,既大幅度提高编码速度,又严格控制码率的增长。 6. Taken together, the present invention has significant characteristics may macroblock directly used to determine the appropriate method, the best coding mode obtained in advance, while not obvious macroblock using the original method, so that due account of all the case, not only greatly improve the encoding speed, and strictly control the rate of growth. 本发明的技术方案流程图参见图4、图5。 Aspect of the present invention. Referring to the flowchart of FIG 4, FIG 5. 图4是R264采用的编码模式选择方法,图5是本发明的方法流程图。 FIG 4 is a coding mode selection method using R264, FIG 5 is a flowchart of a method according to the present invention. 本发明的方法包括下述步骤:1) 从视频帧中提取需要编码的宏块的亮度值;2) 用绝对差值和,即SAD值来表征宏块的相关性,&41)/_代表宏块帧内预测的残差值,以此来表征宏块的空间相关性,&4/),*代表宏块帧间预测的残差值,以此来表征宏块的时间相关性;3) &4"自的计算为下面公式:<formula>formula see original document page 7</formula>式中,^A自代表宏块帧内预测的残差值,及w是视频帧中编码宏块对应的像素亮度值,x、 y是宏块在视频帧中的位置,m、 n是宏块中的像素位置; 用编码宏块的周围相邻像素作为预测像素,经过垂直和水平预测,得到最终的残差值;<formula>formula see original document page 7</formula>的计算为下面公式:<formula>formula see original document page 7</formula>式中,&41),_代表宏块帧间预测的残差值,i^表示当前宏块的像素亮度值,^表示前一帧中对应位置宏块 The method of the present invention comprises the following steps: 1) extracting from the video frame to be encoded macroblock luminance value; 2) and with the absolute difference, i.e., to characterize the correlation value SAD macroblock, & 41) / _ represents Macro block intra prediction residual values, in order to characterize the spatial correlation of the macroblocks, & 4 /), * represents the residual value of the inter-prediction macroblock, a macroblock in order to characterize the time correlation; 3) & 4 "is calculated from the following formula: <formula> formula see original document page 7 </ formula> where, ^ a residues represent a difference from the macroblock intra prediction, and w is the video frame coded macroblocks corresponding to the pixels the luminance value, x, y is located macroblock in a video frame, m, n is the pixel position in the macroblock; with surrounding macroblocks coded neighboring pixels as a predicted pixel through the vertical and horizontal prediction, the final residue to give difference; <formula> formula see original document page 7 </ formula> is calculated as the following equation: <formula> formula see original document page 7 </ formula> wherein, & 41), _ represents an inter prediction residual macroblock difference, I ^ represents a pixel intensity values ​​of the current macroblock, ^ represents the corresponding position in the previous frame macroblock 的像素亮度值,x、 y代表宏块在视频帧中的位置,m、 n代表宏块中像素的位置,经过相减得到预测的残差值;5) 在得到相应的&4Z),自和&4A^后,通过比较二者的大小来决定宏块的预测模式,分别针对&4D^。和引入比较因子《和,实际分别比较的是&4D— 乘以系数"后与&4Z),_的差值和&4Dtora与乘以系数/?后的差值;6) 最终的结果为比较以下两个公式的大小:<formula>formula see original document page 7</formula>上两式中,D砍衡量是否舍弃帧内预测模式,"值取范围为[O, l]间的数,如果"砍值大于零,则说明帧内预测模式的残差值大于帧间预测模式的残差值,舍弃帧内预测模式,该宏块应该采用帧间编码模式;同样"&用来衡量是否舍弃帧间预测模式,"值取值范围为[O, 1],如果D眠小于零,则说明帧间预测模式的残差值大于帧内预测模式的残差值,证明宏块的帧内相关性大于帧 Position of the pixel luminance value, x, y represent macroblocks in a video frame, m, position n represents the pixel macroblock, after the subtraction to give residual values ​​predicted; 5) to give the corresponding & 4Z), from and after & 4A ^, determined by the size of the two comparison prediction mode for the macroblock, respectively, for & 4D ^. and incorporated comparison factor "and actually were compared is & 4D- multiplied by the coefficient" post and & 4Z), _ difference ? multiplied by the difference between the coefficient and & 4Dtora / post; 6) the final result of the magnitude comparison of two formulas: the two formula <formula> formula see original document page 7 </ formula>, D is cut to measure discard an intra prediction mode, "the value taken in the range [O, l] between the number of, if the" cut value greater than zero, then the residual value residual values ​​is greater than the intra-prediction mode of the inter prediction mode, the intra prediction discarded mode, the macro block should be inter-frame coding mode; similarly "& discard used to measure the inter prediction mode," the value in the range [O, 1], sleep if D is less than zero, then the inter prediction mode residual values ​​is greater than the residual values ​​of the intra prediction mode, intra-proof frame correlation is greater than a macroblock 间相关性,该宏块应该采用帧内编码模式;7) 通过上一步判断,如果得到的结果是帧内残差小,采用帧内预测的方法编码,则对残差值做变化、量化、熵编码;8) 通过步骤6的判断,如果得到的结果是帧间残差小,采用运动估计与补偿的编码模式,则对宏块做运动估计,得到最佳匹配块,求相应的差值,对该值做变换、量化、熵编码;输出最终的压縮码流,同时保存相关的编码信息。 Correlation between the macroblock to be intra-frame coding mode; 7) by the determination step, the result is obtained if the residual small frame, intra prediction coding method, the residual values ​​do change, quantization, entropy encoding; 8) is determined by the step 6, if the result obtained is an inter small residuals, using motion estimation and compensation mode encoding, the macroblock motion estimation done, to give the best matching block, find the corresponding difference , this value do transform, quantization, entropy encoding; final compressed output bit stream, while preserving the relevant information is encoded. 本发明的原理为:通过对实际情况的考察,发现对于一般的视频序列, 除非是镜头大幅度的切换,否则在两帧之间变化的内容是非常小的,通过运动估计与补偿的算法,可以找到相对最佳的匹配块,预测残差已经非常小了, 比通过空间相邻的宏块的预测要更好一些。 The principles of the present invention is as follows: by examining the actual situation found for the general video sequence, unless substantial cutaways, or content changes between the two is very small, by the motion estimation and compensation algorithm, find the best matching block relative prediction residual has been very small, than predicted by the macro block adjacent to the space better. 在对实际的视频测试序列进行一定的测试后,发现在实际情况当中,P帧中采用帧内编码模式即Intra4X4和Intral6X16的宏块非常少。 After the actual video test sequences of certain tested in the actual case, the intraframe coding mode, i.e., P frame macroblock uses very little of Intra4X4 and Intral6X16. 在本发明中,考虑到不同的编码模式,对应不同特征的宏块,利用宏块自身的空间相关性和时间相关性来判断宏块采用的编码模式,对于时间相关性大的宏块直接采用运动估计与补偿的模式,对于空间相关性大的宏块直接采用帧内预测的模式,避免了原来方法中的穷举式方法。 In the present invention, in consideration of different coding modes, corresponding to different characteristics of the macroblock by macroblock own spatial correlation and time correlation of the determined macroblock coding modes used for time-dependent large macroblocks used directly motion estimation and compensation mode, direct mode for the macroblock using the intra prediction large spatial correlation, the original method to avoid exhaustive method. 实验结果表明,本方法与H.264中采用的帧间编码方法相比,在码流只有少许增加,视频质量基本不变的情况下,大幅度地提高了编码速度。 Experimental results show that the present method compared to inter-frame coding method employed in H.264, only a few streams at increased substantially constant video quality, the encoding speed will increase greatly. 附图说明图1是宏块的分割方法;图2是垂直预测法示意图;图3是水平预测法示意图;图4是原标准方法中采用的方法的流程图;图5是本发明采用的方法的流程图;具体实施方式在实际的使用当中,首先是读入某个视频序列,输入的视频序列需要是YUV格式的,从计算机中读入视频数据后,本方法是针对序列中的亮度分量来完成的,所以需要从其中提取亮度分量,编码器将调用本发明中提到的算法来完成具体的编码工作。 BRIEF DESCRIPTION OF DRAWINGS FIG. 1 is a method of dividing a macroblock; FIG. 2 is a schematic vertical prediction method; FIG. 3 is a schematic horizontal prediction method; FIG. 4 is a flowchart of a method employed in the original standard method; FIG. 5 is employed in the method of the present invention the flowchart; DETAILED DESCRIPTION in the practical use of them, is first read into a video sequence, the video sequence needs to be input YUV format, the video data is read from the computer, the method is directed to the luminance component of the sequence to complete, from which it is necessary to extract the luminance component, the encoder of the present invention mentioned call algorithm to accomplish specific coding. 具体实施中,在计算机中完成以下程序:第一步:根据编码器配置文件读入相应的视频序列,按照配置文件中的参数配置编码器。 In particular embodiments, the following procedure is completed in the computer: The first step: The encoder configuration file is read into the corresponding video sequence, the encoder configuration parameters in the configuration file. 第二步:从原始文件中取出亮度分量。 Step two: the luminance component extracted from the original file. 第三步:按顺序取出需要编码的宏块的亮度分量。 The third step: the luminance component of the macroblock to be coded in the order taken. 第四步:对该宏块的亮度值做空间垂直、水平预测,得到相应的帧内预测残差,并且乘以比较系数"。第五步:对该宏块的亮度值做帧间预测,用当前宏块的值减去前一帧中相应位置宏块的值,得到相应的帧间预测残差,并且乘以比较系数P。第六步:比较乘系数后的宏块帧内预测残差和帧间预测残差的大小。第七步:如果帧内残差小,则采用帧内预测的方法编码,对残差值做变化、量化、熵编码。第八步:如果帧间残差小,则采用帧间的方法编码,对宏块做运动估计, 得到最佳匹配块,求相应的差值,对该值做变换、量化、熵编码。第九步:输出得到的最终的压縮码流,同时保存相关的编码信息。为了检验本发明所提出的方法的性能,给出本发明与H.264中采用的方法的比较。在比较中,主要从信噪比、编码时间、压縮码流三个方面比较。在实验中 The fourth step: the luminance value of the macroblock to do vertical space, horizontal prediction, an intra prediction residual to obtain the corresponding, comparison and multiplying coefficient "Step 5: the luminance value of the inter prediction macroblock do, macroblock intra prediction coefficient by comparing the residues: subtracting a value corresponding to the value of the current position of the macroblock of the macroblock in the previous frame, to give the corresponding inter prediction residual, and multiplied by coefficient comparison sixth step P. and the difference between the size of the inter prediction residuals seventh step: If the intra small residual, the intra prediction encoding method, residual values ​​do change, quantization, entropy encoding eighth step: If the residual inter small difference, the interframe coding method, the motion estimation done macroblock, the best matching block, find the corresponding difference values ​​do the conversion, quantization, entropy encoding step 9: the output obtained final compressed stream, while preserving the relevant encoded information. in order to test the performance of the proposed method of the present invention, the method of the present invention give a more employed in H.264. in comparison, mainly from the signal-encoding time , three comparison compressed stream in the experiment 采用JVT公布的编码校验模型jm5.0进行试验,实验条件如下-主机为P4 2.8CPU, 256M内存,编码30帧,第一帧采用I帧编码,其余的采用P帧编码。实验结果如表l,从实验结果中可以看出,本发明的方法与原方法比较,在视频质量方面,也就是信噪比方面几乎没有任何影响。平均码率只增加1.21%,速度的提高平均有24%。从结果中可以看出,在码率只有少许增加的情况下,速度有相当大的提高,证明了本发明的有效性。表l:实验结果比较序列名 PSNRY PSNRU PSNRV 速度提高(%) 码率增Trevor 升高0.01 dB 升高0.02 不变 12.69 1.73<table>table see original document page 10</column></row> <table> Jm5.0 JVT coding model using published check test, the following experimental conditions - host P4 2.8CPU, 256M memory, code 30, encoding a first frame using I frame, P-frame coding using the remaining test results are shown in Table. L, can be seen from the experimental results, the method of the present invention is the original method of comparison, in terms of video quality, signal to noise ratio is little or no effect. the average rate increased by only 1.21%, an average speed increase of 24% As can be seen from the results, in the case where only a small increase rate, velocity considerably improved, the effectiveness of the present invention is demonstrated in table l:. the results were compared sequences PSNRY PSNRU PSNRV speed (%) code Trevor rate increased by 0.01 dB 0.02 constant elevated 12.69 1.73 <table> table see original document page 10 </ column> </ row> <table>

Claims (1)

1. 基于时空相关性的视频压缩方法,其特征在于:根据视频数据自身的特点,利用其自身的空间、时间相关性来对帧间宏块进行预测、编码;在视频编码标准H.264中采用的帧间宏块压缩方法的基础上,利用空间相关性和时间相关性提前判断宏块是应该采用帧内编码模式还是运动估计与补偿的编码模式;包括下述步骤: 1)从视频帧中提取需要编码的宏块的亮度值; 2)用绝对差值和,即SAD值来表征宏块的相关性,SADIntra代表宏块帧内预测的残差值,以此来表征宏块的空间相关性,SADInter代表宏块帧间预测的残差值,以此来表征宏块的时间相关性; 3)SADIntra的计算为下面公式: 1. The space-time correlation-based video compression methods, which is characterized in that: its own characteristics, using its own spatial and time correlation of the inter-predicted macroblocks, encoding the video data; in the H.264 video coding standard the inter macroblock basis using the compression method using a spatial correlation and a temporal correlation is determined in advance to be a macroblock is coded using intra-mode or the motion estimation and compensation encoding mode; comprising the steps of: 1) from the video frame luminance values ​​extracted macroblock to be encoded; 2) and with the absolute difference, i.e. the macroblock SAD value characterized correlation representative SADIntra intra prediction residuals macroblocks, in order to characterize the spatial macroblock correlation, SADInter residual value representative of the inter-prediction macroblock, a macroblock in order to characterize the time correlation; 3) SADIntra calculated as the following formula:
CN 200610080881 2006-05-19 2006-05-19 Video interframe compression method based on space-time correlation CN100409690C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN 200610080881 CN100409690C (en) 2006-05-19 2006-05-19 Video interframe compression method based on space-time correlation

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN 200610080881 CN100409690C (en) 2006-05-19 2006-05-19 Video interframe compression method based on space-time correlation

Publications (2)

Publication Number Publication Date
CN1845608A CN1845608A (en) 2006-10-11
CN100409690C true CN100409690C (en) 2008-08-06

Family

ID=37064517

Family Applications (1)

Application Number Title Priority Date Filing Date
CN 200610080881 CN100409690C (en) 2006-05-19 2006-05-19 Video interframe compression method based on space-time correlation

Country Status (1)

Country Link
CN (1) CN100409690C (en)

Families Citing this family (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101453559B (en) 2007-12-04 2010-12-08 瑞昱半导体股份有限公司 Noise detection method and apparatus for video signal
CN102843142B (en) * 2011-06-23 2013-07-31 深圳市国微电子有限公司 Compression and decompression processing method and system of configuration data stream for programmable logic device
CN102510496B (en) * 2011-10-14 2013-11-13 北京工业大学 Quick size reduction transcoding method based on region of interest
CN103079068A (en) * 2013-01-14 2013-05-01 天津大学 Interframe mode judging method based on relevance between motion intensity and space domain

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1302510A (en) 1999-04-30 2001-07-04 皇家菲利浦电子有限公司 Video encoding method with selection of B-frame encoding mode
US20020122487A1 (en) 1998-08-26 2002-09-05 Yasuko Matsumura Motion image encoder
CN1457196A (en) 2003-04-25 2003-11-19 北京工业大学 Video encoding method based on prediction time and space domain conerent movement vectors
CN1585495A (en) 2004-06-11 2005-02-23 上海大学 Quick selection of prediction modes in H.264/AVC frame
CN1774070A (en) 2004-11-09 2006-05-17 C&S技术有限公司 Motion estimation method using adaptive mode decision

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20020122487A1 (en) 1998-08-26 2002-09-05 Yasuko Matsumura Motion image encoder
CN1302510A (en) 1999-04-30 2001-07-04 皇家菲利浦电子有限公司 Video encoding method with selection of B-frame encoding mode
CN1457196A (en) 2003-04-25 2003-11-19 北京工业大学 Video encoding method based on prediction time and space domain conerent movement vectors
CN1585495A (en) 2004-06-11 2005-02-23 上海大学 Quick selection of prediction modes in H.264/AVC frame
CN1774070A (en) 2004-11-09 2006-05-17 C&S技术有限公司 Motion estimation method using adaptive mode decision

Also Published As

Publication number Publication date
CN1845608A (en) 2006-10-11

Similar Documents

Publication Publication Date Title
CN101448162B (en) Method for processing video image
CN1204751C (en) Coded data converting method, recoding method, recoding system and data recording medium
CN101273641B (en) Dynamic image encoding device and dynamic image decoding device
CN101939992B (en) Reference selection for video interpolation or extrapolation
CN101924935B (en) Advanced bi-directional predictive coding of interlaced video
CN100361536C (en) Device and method for predicting and encoding image and recording medium
CN100518321C (en) Deblocking control method, and multilayer video encoder/decoder using the same
CN101138250B (en) Adaptive intra-frame refresh for digital video encoding
CN1843037B (en) Method and apparatus for encoding hybrid intra-inter coded blocks
CN1981531B (en) Method and apparatus to construct bi-directional predicted frames for temporal scalability
CN101584219B (en) A method and apparatus for encoding and decoding based on intra prediction
CN100536573C (en) Inframe prediction method used for video frequency coding
CN101175209B (en) Adaptive de-blocking filtering method for mpeg video decoder
CN1905679B (en) Image encoding device, image encoding method, image decoding device, image decoding method, and communication device
CN101352046B (en) Image encoding/decoding method and apparatus
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
CN100579224C (en) Method and apparatus for lossless video encoding and decoding
CN100387059C (en) Picture decoding method
CN1906945B (en) Method and apparatus for scalable video encoding and decoding
CN100568976C (en) Method and device for intra prediction coding and decoding of image
CN1981535B (en) Method and apparatus for weighted prediction in predictive frames
CN101711481B (en) Method and apparatus for video coding using prediction data refinement
CN1316830C (en) Moving picture coding apparatus, moving picture decoding apparatus, moving picture coding method, and moving picture decoding method
CN101273638B (en) Image processing device, image processing method
CN100566426C (en) Method and apparatus for encoding and decoding of video

Legal Events

Date Code Title Description
C06 Publication
C10 Request of examination as to substance
C14 Granted
C17 Cessation of patent right