WO2005027520A1 - Bi-directional predicting method for video coding/decoding - Google Patents

Bi-directional predicting method for video coding/decoding Download PDF

Info

Publication number
WO2005027520A1
WO2005027520A1 PCT/CN2004/000735 CN2004000735W WO2005027520A1 WO 2005027520 A1 WO2005027520 A1 WO 2005027520A1 CN 2004000735 W CN2004000735 W CN 2004000735W WO 2005027520 A1 WO2005027520 A1 WO 2005027520A1
Authority
WO
WIPO (PCT)
Prior art keywords
motion vector
block
current
backward
candidate
Prior art date
Application number
PCT/CN2004/000735
Other languages
English (en)
French (fr)
Inventor
Xiangyang Ji
Wen Gao
Debin Zhao
Yan Lu
Siwei Ma
Honggang Qi
Original Assignee
Institute Of Computing Technology Chinese Academy Of Sciences
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 Institute Of Computing Technology Chinese Academy Of Sciences filed Critical Institute Of Computing Technology Chinese Academy Of Sciences
Priority to EP04738333.6A priority Critical patent/EP1672926B1/en
Priority to BRPI0413945-3A priority patent/BRPI0413945B1/pt
Priority to JP2006525606A priority patent/JP4755095B2/ja
Priority to US10/571,659 priority patent/US8005144B2/en
Publication of WO2005027520A1 publication Critical patent/WO2005027520A1/zh

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51—Motion estimation or motion compensation
    • H04N19/577—Motion compensation with bidirectional frame interpolation, i.e. using B-pictures
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51—Motion estimation or motion compensation
    • H04N19/513—Processing of motion vectors
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51—Motion estimation or motion compensation
    • H04N19/56—Motion estimation with initialisation of the vector search, e.g. estimating a good candidate to initiate a search
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/50—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding
    • H04N19/503—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using predictive coding involving temporal prediction
    • H04N19/51—Motion estimation or motion compensation
    • H04N19/58—Motion compensation with long-term prediction, i.e. the reference frame for a current frame not being the temporally closest one
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
    • H04N19/60—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding
    • H04N19/61—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using transform coding in combination with predictive coding

Definitions

  • the present invention relates to a bidirectional prediction method for video encoding, in particular to a bidirectional prediction method for compressing a video; and belongs to the technical field of video encoding and decoding. Background technique
  • Digital TV, next-generation mobile communications, broadband network communications, and home consumer electronics are booming high-tech industrial clusters. Their common technologies are focused on multimedia information processing technologies with video and audio as the main content, especially data compression technology. Efficient video codec technology is the key to high-quality, low-cost multimedia data storage and transmission.
  • commonly used encoding methods include predictive encoding, orthogonal transform encoding, vector quantization encoding, etc. These methods are all based on signal processing theory, and are also commonly referred to as the first generation encoding technology.
  • the more popular international standards for image coding are based on this coding theory, using a coding method based on block matching motion compensation, discrete cosine transform, and quantization.
  • Typical are the Motion Picture Experts Group (MPEG)-1, MPEG-2 and MPEG-4, etc., introduced by the International Organization for Standardization / International Electrotechnical Commission First Joint Technical Group (IS0 / IEC JTC1) International standards, and the H. 26x series of recommendations from the International Telecommunication Union (ITU-T). These video coding standards are widely used in industry.
  • hybrid Video Coding Hybrid Video Coding
  • the main function of the prediction module is to use the already encoded and reconstructed image to predict the current image to be encoded (inter prediction), or to use the already encoded and reconstructed image block in the image to predict the currently encoded image block (intra frame) Prediction);
  • the main function of the transformation module is to transform the input image block to another space, so that the energy of the input signal is concentrated on the low-frequency transform coefficients as much as possible, thereby reducing the correlation between the elements in the image block, which is conducive to compression
  • the main function of the quantization module is to map the transformed coefficients to a finite element set that is conducive to coding.
  • the main function of the information entropy coding module is to represent the quantized transformation coefficients with variable length codes according to statistical rules.
  • the video decoding system contains similar modules, which are mainly input code streams The decoded image is reconstructed through processes such as entropy decoding, inverse quantization, and inverse transform.
  • video encoding and decoding systems usually include some auxiliary encoding tools, which will also contribute to the overall system's encoding performance (compression ratio).
  • the main function of motion-compensated prediction is to eliminate temporal redundancy of video sequences.
  • the video encoding process is the process of encoding each frame of the video sequence.
  • the prediction module is the one that completes this function.
  • the commonly used video encoding system encodes each frame of image by using image blocks as the basic unit. When encoding each frame of image, it can be divided into intra-frame (I-frame) encoding, prediction (P-frame) encoding, and bi-directional prediction (B-frame) encoding. Generally, when encoding, I-frame, P-frame, and B-frame encoding are interspersed, for example, in the order of IBBPBBP.
  • the introduction of B frames can effectively solve the "occlusion problem" caused by different moving directions or speeds between moving objects or between objects and background.
  • the encoding of B frames can make the encoding compression efficiency reach a code rate above 200: 1.
  • the coding of the image blocks in the B frame needs to include four modes: direct, forward prediction, backward prediction, and bi-directional prediction. Since the B-frame technology needs to perform forward and backward motion estimation simultaneously, it requires a high computational complexity, and additional identification information is introduced in order to distinguish forward and backward motion vectors.
  • the technical problem to be solved by the present invention is to propose a bidirectional prediction method for video encoding, which can effectively reduce the number of motion vectors required for encoding, and does not substantially increase the complexity of searching for matching blocks at the encoding end.
  • the method for bidirectional prediction at the encoding end of video encoding includes the following steps:
  • Step 10 For each image block of the current B frame, use a forward prediction mode to obtain a forward motion vector candidate of the current image block from the forward reference image;
  • Step 20 Use the candidate forward motion vector of the current image block obtained in step 10 to calculate a candidate backward motion vector, and obtain the candidate forward motion vector and candidate backward motion vector required for bidirectional prediction;
  • Step 30 Use the candidate forward motion vector and candidate backward motion vector of the current image block obtained in step 20 to obtain a candidate bidirectional prediction reference block through a bidirectional prediction method;
  • Step 40 Within the given search range and / or before the matching value is less than or equal to the preset matching threshold, continuously set new reference blocks, and repeat the previous three steps to select the best matching block. ;
  • Step 50 Code the forward motion vector, backward motion vector, and block residual of the image block determined by the optimal reference block into the code stream.
  • the decoding-side bidirectional prediction method for video encoding includes the following steps:
  • Step 21 Decode from the code stream to obtain a forward motion vector.
  • Step 31 When the forward motion vector is obtained in step 21, calculate a backward motion vector, and obtain a forward motion vector and a backward motion vector required for bidirectional prediction.
  • Step 41 Use the forward motion vector and backward motion vector of the current image block obtained in step 31 to obtain a final bidirectional prediction reference block through a bidirectional prediction method;
  • Step 51 Combine the prediction reference block obtained in step 41 and the block residual obtained by decoding correspondingly in the code stream to form a current block image block.
  • the bidirectional prediction method for video coding of the present invention encodes only one motion vector, and the other motion vector is obtained through calculation, and achieves the purpose of bidirectional prediction. It is also called a single motion vector bidirectional prediction method.
  • the method of the present invention is basically Without increasing the complexity of searching for matching blocks at the coding end, the amount of coding for the motion vector can be greatly saved, and the method of the present invention can also more realistically reflect the motion of the object in the video, obtain more accurate motion vector prediction, and forward
  • the combination of predictive coding and backward predictive coding can be used to implement a new type of predictive coding.
  • FIG. 1 is a schematic diagram of a backward motion vector derivation process in frame coding
  • FIG. 2 is a schematic diagram of a backward motion vector derivation process in a field coding in an odd field or an even field when a motion vector of a corresponding block of a backward reference field points to a field earlier in the time domain than a current field;
  • FIG. 3 is a schematic diagram of a backward motion vector derivation process in field coding in an even field when a motion vector of a block corresponding to a backward reference field points to an odd field corresponding to the same frame as the even field;
  • FIG. 4 is a bidirectional prediction flowchart of the forward motion vector obtained at the encoding end to obtain the forward motion vector to calculate the backward motion vector and finally to obtain the optimal matching block;
  • Figure 5 shows how the de-encoding end derives the backward motion vector from the forward motion vector obtained from the code stream, and finally reconstructs an image block through bidirectional prediction compensation.
  • the bidirectional prediction encoding method includes the following steps:
  • Step 10 For each image block of the current B frame, use a forward prediction mode to obtain a forward motion vector of the current image block candidate from the forward reference image.
  • the forward prediction mode is specifically:
  • Step 101 if the first packet to the reference pictures ⁇ Shu given reference block, perform Step 102; otherwise, executing step 103;
  • Step 102 The position of the reference block set in the forward reference picture in the forward reference picture is different from the position of the current image block in the B frame in the current picture, and the obtained vector is a candidate forward motion vector. End step 1 0;
  • Step 1 Select the image block in the forward reference picture that is at the same position as the current image block in the B frame as the reference block set in the forward reference picture, and perform step 102.
  • the difference will not be particularly great.
  • the same point as the reference picture is selected as the reference point.
  • the candidate forward motion vector is 0, and there is no position change between the two.
  • the reference point is changed in step 40, and the candidate forward motion vector is no longer zero.
  • Step 20 Use the candidate forward motion vector of the current image block obtained in step 10 to calculate the candidate backward motion vector, and obtain the candidate forward motion vector and candidate backward motion vector required for bidirectional prediction; specifically:
  • the candidate forward motion vector and candidate backward motion vector of the current block can be calculated by the following formula:
  • TD B is the time domain distance between the current B frame and the forward reference frame
  • TD D A is the time domain distance between the backward reference frame and the forward reference frame
  • CMV F and CMV B are the candidate forwards of the current block of the corresponding B frame, respectively.
  • the motion vector and candidate backward motion vector are shown in Figure 1.
  • the candidate forward motion vector and candidate backward motion vector of the current block can be calculated by the following formula:
  • CMV BJ -(- ⁇ - ⁇ X CMV F
  • TD B is the distance in the time domain between the current picture and the forward reference picture
  • TD D is the distance in the time domain between the forward reference picture and the backward reference picture
  • CMV F and CMV B are the corresponding B derived respectively.
  • the backward motion vector of the current block is derived in accordance with the odd field.
  • the candidate forward motion vector and candidate backward motion vector of the current block are derived as follows: MV B CMV F ⁇ X
  • TD B is the distance in the time domain between the current picture and the forward reference picture
  • TD D is the distance in the time domain between the forward reference picture and the backward reference picture
  • CMV F and CMV B are the corresponding B-frame currents derived from the current
  • the candidate forward motion vectors and candidate backward motion vectors for the block are shown in Figure 3.
  • Step 30 Use the candidate forward motion vector and candidate backward motion vector of the current image block obtained in step 20 to obtain the final bidirectional prediction reference block through a bidirectional prediction method; that is, the candidate forward motion vector and the candidate backward motion vector are The pixels corresponding to the two prediction reference blocks are averaged to obtain the final bidirectional prediction reference block.
  • Step 40 Within the given search range and / or before the matching value is less than or equal to a preset matching threshold, continuously set new reference blocks, repeat the previous three steps, and finally select the optimal matching block. ;
  • the search range is a certain area centered on the reference block in the reference picture that is the same as the current block position of the B frame.
  • the size of the search area varies with different requirements for image quality. The larger the search area, the The more accurate the reference block is, the larger the search area can be throughout the entire reference picture.
  • the bidirectional prediction reference block with the smallest sum (indicated by SAD) of the absolute value of the difference between the bidirectional prediction reference block obtained from the reference block in the entire search range and the corresponding pixel in the B frame is the best matching block.
  • the matching value is the sum of the absolute value of the difference between the bidirectional prediction reference block and the corresponding pixel of the B-frame current block, SAD.
  • the matching threshold value is a preset matching value. If the matching value is less than or equal to the matching threshold value, the reference block at this time is the optimal reference block. In a certain order, the current reference block is used as the base point to calculate the matching value of the reference block from near to far.
  • the method of calculating SAD is adopted to represent the difference between the bidirectional prediction reference block and the current block of frame B.
  • Other methods such as calculating the variance of corresponding pixels, are not as intuitive and efficient as SAD method.
  • the matching value is calculated from near to far within the set area; this way, the search range can be determined as needed; and it is not necessary to traverse the entire area.
  • the search range is the most efficient.
  • Step 50 Code the forward motion vector, backward motion vector, and block residual of the image block determined by the optimal reference block into the code stream.
  • the difference between the bidirectional prediction reference block determined by the optimal reference block and the corresponding pixel of the current block in the B frame can be directly encoded, or the difference sequence between the optimal reference block and the corresponding pixel in the B frame current block can be directly encoded. Compression is good for transmission.
  • the Chinese-oriented prediction method for video encoding according to the embodiment of the present invention includes the following steps:
  • the bidirectional prediction decoding method includes the following steps:
  • Step 21 Decode from the code stream to obtain a forward motion vector.
  • Step 31 When the forward motion vector is obtained in step 21, calculate a backward motion vector, and obtain a forward motion vector and a backward motion vector required for the Chinese direction prediction;
  • Step 41 Use the forward motion vector and backward motion vector of the current image block obtained in step 31 to obtain a final bidirectional prediction reference block through a bidirectional prediction method;
  • Step 51 Combine the prediction reference block obtained in step 41 and the block residual obtained by decoding correspondingly in the code stream to form a current block image.
  • step 31 the step of calculating the backward motion vector is as follows:
  • Step 310 Determine the current image mode. If it is a frame encoding mode, go to step 311; if it is a field encoding mode, determine whether it is an odd field or an even field; if it is an odd field, go to step 312; if it is an even field, go to step 31 3;
  • Step 311 Use the following formula to calculate the backward motion vector:
  • TD B is the distance in the time domain between the current picture and the forward reference picture.
  • TD D is the distance in the time domain between the forward reference picture and the backward reference picture.
  • MV F and MV B are the front of the current block of the corresponding B frame. Motion vector and backward motion vector; end step 31;
  • Step 312 Calculate and obtain the backward motion vector by the following formula:
  • TD B is the distance in the time domain between the current picture and the forward reference picture
  • TD D is the distance in the time domain between the forward reference picture and the backward reference picture
  • ⁇ ⁇ and MV B are the corresponding B derived respectively
  • the forward motion vector and backward motion vector of the current block of the frame; the value of the subscript i is determined according to the parity mode, 0 for odd mode, and 1 for even mode. End step 31;
  • Step 313 When the motion vector of the block corresponding to the backward reference field points to a field earlier than the current field in the time domain, step 312 is performed; when the motion vector of the block corresponding to the backward reference field points to the same frame as the even field For odd fields, the backward motion vector is calculated by the following formula:
  • TD B is the distance in the time domain between the current picture and the forward reference picture
  • TD D is the distance in the time domain between the forward reference picture and the backward reference picture
  • MV F and MV B are the corresponding B-frame currents derived from the current Forward motion vector and backward motion vector of the block
  • step 41 the specific process of the bidirectional prediction method is: averaging the pixels corresponding to the two prediction reference blocks pointed by the forward motion vector and the backward motion vector to obtain the final bidirectional prediction reference block.
  • the decoding process is relatively simple. After obtaining the forward motion vector from the code stream, the backward motion vector is directly calculated, and then the bidirectional prediction reference block is combined with the block residual to obtain the image before encoding. This can be considered as encoding. Reverse process of process.

Landscapes

  • Engineering & Computer Science (AREA)
  • Multimedia (AREA)
  • Signal Processing (AREA)
  • Compression Or Coding Systems Of Tv Signals (AREA)

Description

用于视频编码的编码端 /解码端汉向预测方法 技术领域
本发明涉及一种用于视频编码的双向预测方法， 尤其是一种用于对视频进 行压缩的双向预测方法； 属于视频编解码技术领域。 背景技术
数字电视、 新一代移动通信、 宽带网络通信和家庭消费电子这些蓬勃发展 的高技术产业群， 其共性技术集中在以视音频为主要内容的多媒体信息处理技 术， 特别是数据压缩技术上。 高效的视频编解码技术是实现高质量、 低成本多 媒体数据存储与传输关键。 目前常用的编码方法有预测编码、 正交变换编码、 向量量化编码等等， 这些方法都是基于信号处理理论的， 通常也称为第一代编 码技术。 现在比较流行的图像编码国际标准都是基于这种编码理论， 采用的是 基于块匹配的运动补偿、 离散余弦变换和量化相结合的编码方法。 典型的有国 际标准化组织 /国际电工技术委员会第一联合技术组（IS0/IEC JTC1 )推出的 运动图象专家组（Mot ion Picture Experts Group, 简称 MPEG) - 1， MPEG - 2 和 MPEG-4等国际标准， 以及国际电信联盟（ITU- T )提出的 H. 26x系列推荐。 这 些视频编码标准在工业界得到了广泛应用。
这些视频编码标准都采用了混合视频编码（Hybr id Video Coding ) 策略， 通常包括： 预测、 变换、 量化和信息熵编码等四个主要模块。 预测模块的主要 功能时利用已经编码并重建的图像对当前要编码的图像进行预测 （帧间预测）， 或者利用图像中已经编码并重建的图像块对当前要编码的图像块进行预测 （帧 内预测）； 变换模块的主要功能是将输入的图像块变换到另外一个空间， 使输入 信号的能量尽可能地集中在低频变换系数上， 进而降低图像块内元素之间的相 关性， 有利于压缩； 量化模块的主要功能是将变换的系数映射到一个有利于编 码的有限元素集上； 信息熵编码模块的主要功能是根据统计规律， 将量化后的 变换系数用变长码表示。 视频解码系统包含相似的模块， 主要是将输入的码流 通过熵解码、 反量化、 反变换等过程重建解码图像。 除了上述模块， 视频编解 码系统中通常还包含一些辅助的编码工具， 这些工具也会对整个系统的编码性 能 （压缩比）做出贡献。
视频编码的大部分编码效率来自于基于运动补偿的预测。 基于运动补偿的 预测的主要功能是消除视频序列在时间上的冗余。 视频编码过程就是对视频序 列的每一帧图像进行编码的过程， 完成这一功能的就是预测模块。 常用的视频 编码系统对每一帧图像的编码是以图像块为基本单元进行的。 在编码每一帧图 像时， 又可以分为帧内 （ I帧）编码、 预测 （ P帧）编码和双向预测（ B帧）编 码等情况。 一般来说， 编码时， I帧、 P帧和 B帧编码是穿插进行的， 比如按照 IBBPBBP的顺序。
B 帧的引入可以有效地解决运动物体之间或物体与背景之间由于不同的运 动方向或运动速率而引起的 "遮挡问题"。 B帧的编码可以使得编码压缩效率达 到 200: 1以上的码率。 对 B帧中的图像块进行编码需要包括： 直接（Direct )、 前向预测 ( Forward Predict ion ), 后向预测 ( Backward Predict ion )和双向 预测（Bi-direct ional Predict ion )四种模式。 由于 B帧技术需要同时进^ "前 向与后向的运动估计， 因此需要较高的运算复杂度， 同时为了区分前后向运动 矢量要引入额外的标识信息。
在现有视频编码系统中， B 帧通常存在双向预测的运动模式， 因为这种模 式可以有效地消除由于图片之间的旋转、 亮度的变化、 噪声等引起的帧间预测 不准确性。 但是， 与此同时也需要对更多的运动矢量进行编码， 因此， 运动矢 量编码的比特数在整个编码中的比重经常要大于 30%。
因此， 如果有一种方法， 在保持较好地双向预测性能前提下， 降低对运动 矢量的编码， 将会有效地提高编码的压缩率， 特别是对于低码率视频传输的应 用， 降低运动矢量编码所需要的比特数更有意义。 发明内容 本发明所要解决的技术问题在于提出一种用于视频编码的双向预测方法， 可以有效地降低所需编码的运动矢量数量， 而且基本上不增加编码端搜索匹配 块的复杂度。
本发明的技术方案如下所述：
用于视频编码的编码端的双向预测的方法包括如下步骤：
步骤 1 0、 对当前 B帧的每一图像块， 采用前向预测模式， 从前向参考图像 中获得当前图像块候选前向运动矢量；
步骤 20、 利用步骤 1 0得到的当前图像块的侯选前向运动矢量， 计算得到 候选后向运动矢量， 获得进行双向预测所需的侯选前向运动矢量和侯选后向运 动矢量；
步骤 30、 利用步骤 20得到的当前图像块的侯选前向运动矢量和侯选后向 运动矢量， 通过双向预测方法得到候选双向预测参考块；
步骤 40、 在给定的搜索范围之内和 /或在匹配值小于或等于预先设定的匹 配阈值以前， 不断设定新的参考块， 并重复进行前面三个步骤， 选取最佳的匹 配块；
步骤 50、 将该图像块的由最优参考块确定的前向运动矢量、 后向运动矢量 和块残差， 编入码流中。 所述的用于视频编码的解码端双向预测方法， 包括如下步骤：
步骤 21、 从码流中解码， 获得前向运动矢量；
步骤 31、 利用步骤 21 中得到前向运动矢量时， 计算得到后向运动矢量， 得到了双向预测所需的前向运动矢量和后向运动矢量；
步骤 41、 利用步骤 31得到的当前图像块的前向运动矢量和后向运动矢量， 通过双向预测方法， 求得最终的双向预测参考块；
步骤 51、 将步骤 41得到的预测参考块与码流中解码相应的得到的块残差 合并形成当前块图像块。 本发明的用于视频编码的双向预测方法， 只对一个运动矢量进行编码， 另 一个运动矢量通过计算获得， 实现了双向预测的目的， 也称为单运动矢量双向 预测方法， 本发明方法基本上不增加编码端搜索匹配块的复杂度， 能够极大的 节省对运动矢量的编码量， 并且本发明方法也可以更真实地体现视频中物体的 运动， 获得更准确的运动矢量预测， 与前向预测编码， 后向预测编码相结合， 可用实现一种新的预测编码类型。 附图说明
图 1是在帧编码中的后向运动矢量推导过程示意图；
图 2是在场编码中， 奇场或偶场中当后向参考场对应块的运动矢量指向时 域上早于当前场的某一场时， 后向运动矢量推导过程示意图；
图 3是在场编码中， 偶场中， 当后向参考场对应块的运动矢量指向与偶场 属于同一帧相应的奇场时， 在场编码中后向运动矢量推导过程示意图；
图 4是编码端实现运动估计获得前向运动矢量来计算后向运动矢量以及最 终求得最优的匹配块的双向预测流程图；
图 5 是解编码端如何从码流中获得的前向运动矢量来推导后向运动矢量， 并最终通过双向预测补偿来重构某一图像块的过程。 具体实施方式
下面通过附图和实施例对本发明的技术方案做进一步详细说明： 在本发明的实施例中， 假设只有一个前向参考图片与一个后向参考图片对 当前 B帧的图像块在前后参考帧上进行运动估计。
本发明的实施例的用于视频编码的双向预测方法， 所述的双向预测编码方 法， 如图 4所示， 包括如下步骤：
步骤 1 0、 对当前 B帧的每一图像块， 采用前向预测模式， 从前向参考图像 中获得当前图像块候选的前向运动矢量。 所述的前向预测模式具体为：
步骤 101、 如果前向参考图片中包 ^殳定的参考块， 执行步驟 102； 否则， 执行步骤 1 03 ;
步骤 102、 前向参考图片中设定的参考块在前向参考图片中的位置与 B帧 当前图像块在当前图片中的位置作差， 所得到的向量即为候选前向运动矢量； 结束步骤 1 0 ;
步骤 1 03、 选定前向参考图片中与 B帧当前图像块位置相同图像块作为前 向参考图片中设定的参考块， 执行步骤 1 02。
由于前后两帧的图像时间间隔很短， 相差不会特别多， 起初选择参考图片 位置相同点作为参考点， 此时的候选前向运动矢量为 0， 二者之间没有位置改 变； 如果通过下面步骤 40的改变了参考点， 候选前向运动矢量就不再为 0了。
步骤 20、 利用步骤 10得到的当前图像块的侯选前向运动矢量， 计算得到 侯选后向运动矢量， 获得进行双向预测所需的侯选前向运动矢量和侯选后向运 动矢量； 具体为：
帧编码模式：
在这种模式中， 当前块的候选前向运动矢量和候选后向运动矢量可以通过 如下公式计算：
CMVB = -( Do— £ χ CMVF
Β TDB J b
这里 TDB是当前 B帧与前向参考帧的时域距离， TDDA后向参考帧与前向参 考帧的时域距离， CMVF与 CMVB分别是相应 B帧当前块的候选前向运动矢量和候 选后向运动矢量， 见图 1。
场编码方式：
在奇场这种模式中， 当前块的候选前向运动矢量和候选后向运动矢量可以 通过如下公式计算：
TDD , - TDB ,
CMVBJ = - (-^ - ~~ X CMVF 这里 TDB是当前图片与前向参考图片， 在时域上的距离， TDD是前向参考 图片与后向参考图片在时域上的距离， CMVF与 CMVB分别是推导出来的相应 B 帧当前块的候选前向运动矢量和候选后向运动矢量。 下标 i的取值根据奇偶模 式决定， 为奇模式时取 0 , 偶模式时取 1 , 见图 2。
在偶场这种模式中， 当后向参考场对应块的运动矢量指向时域上早于当前 场的某一场时， 当前块的后向运动矢量的推导与奇场的一致。
当后向参考场对应块的运动矢量指向相应的（与偶场属于同一帧）奇场时， 在这种情况下， 当前块的候选前向运动矢量和候选后向运动矢量推导如下： MVB CMVF<X
Figure imgf000008_0001
TDB是当前图片与前向参考图片在时域上的距离， TDD是前向参考图片与后 向参考图片在时域上的距离， CMVF与 CMVB分别是推导出来的相应 B帧当前块的 候选前向运动矢量和候选后向运动矢量， 见图 3。
步骤 30、 利用步骤 20得到的当前图像块的候选前向运动矢量和候选后向 运动矢量， 通过双向预测方法得到最终的双向预测参考块； 即对候选前向运动 矢量和候选后向运动矢量所指的两个预测参考块对应的像素取平均， 求得最终 的双向预测参考块。
步骤 40、 在给定的搜索范围之内和 /或在匹配值小于或等于预先设定的匹 配阈值以前， 不断设定新的参考块， 重复进行前面三个步骤， 最后选取最优的 匹配块；
在步骤 40中，搜索范围为以参考图片中与 B帧当前块位置相同的参考块为 中心的一定区域， 搜索区域的大小由对图像质量的要求不同而不同， 搜索的区 域越大， 得出的参考块越准确， 搜索区域最大可遍及整个参考图片。 全部搜索 范围中参考块得出的双向预测参考块与 B帧当前块对应像素之差即的绝对值的 和 （用 SAD表示）最小的双向预测参考块为最优匹配块。
匹配值为双向预测参考块与 B帧当前块对应像素之差的绝对值的和 SAD， 匹配阈值为预先设定的匹配值， 如果匹配值小于或等于匹配阈值， 此时的参考 块即为最优参考块。 按一定的顺序， 一般以当前参考块为基点由近及远， 来计 算参考块的匹配值。 采用设定匹配阈值的方法， 可以不必完全遍历所有的参考 点， 找到符合要求的参考块， 就结束最优参考块的搜索过程， 效率很高。
在以上的两种方法中， 采用了计算 SAD的方法来表示双向预测参考块与 B 帧当前块的差异， 也可以采用其他的方法， 比如计算对应像素的方差， 但不如 SAD方法直观， 高效。
当然， 可以采用搜索区域和设定匹配阈值相结合的方法， 如图 4所示， 在 设定的区域内由近及远计算匹配值； 这样可以根据需要， 确定搜索范围； 又不 必完全遍历整个搜索范围， 最为高效。
步骤 50、将该图像块的由最优参考块确定的前向运动矢量、后向运动矢量 和块残差， 编入码流中。 块残差为最优参考块确定的双向预测参考块与 B帧当 前块的对应像素的差异， 可以直接编码最优参考块与 B帧当前块对应像素的差 值序列， 或将该差值序列进行压缩， 利于传送。 本发明的实施例的用于视频编码的汉向预测方法， 所述的双向预测解码方 法， 如图 5所示， 包括如下步骤：
所述的双向预测解码方法， 包括如下步骤：
步骤 21、 从码流中解码， 获得前向运动矢量；
步骤 31、 利用步骤 21 中得到前向运动矢量时， 计算得到后向运动矢量， 得到了汉向预测所需的前向运动矢量和后向运动矢量；
步骤 41、 利用步骤 31得到的当前图像块的前向运动矢量和后向运动矢量， 通过双向预测方法， 求得最终的双向预测参考块；
步骤 51、 将步骤 41得到的预测参考块与码流中解码相应的得到的块残差 合并形成当前块图像。 所述的步骤 31中， 计算后向运动矢量的步骤如下：
步骤 310、 判断当前的图像模式， 如果为帧编码模式， 执行步骤 311 ; 如果 为场编码方式， 判断为奇场还是偶场， 如果为奇场， 执行步骤 312; 如果为偶 场， 执行步骤 31 3;
步骤 311、 通过如下的公式， 计算得到后向运动矢量：
TD D_- TD
TD B
TDB是当前图片与前向参考图片在时域上的距离， TDD是前向参考图片与后 向参考图片在时域上的距离， MVF与 MVB分别是相应 B帧当前块的前向运动矢量 和后向运动矢量； 结束步骤 31 ;
步骤 312、 通过如下的公式， 计算得到后向运动矢量：
Figure imgf000010_0001
TDB是当前图片与前向参考图片， 在时域上的距离， TDD是前向参考图片与 后向参考图片在时域上的距离， {^<与 MVB分别是推导出来的相应 B帧当前块的 前向运动矢量和后向运动矢量； 下标 i的取值根据奇偶模式决定， 为奇模式时 取 0 , 偶模式时取 1。 结束步骤 31 ;
步骤 313、 当后向参考场对应块的运动矢量指向时域上早于当前场的某一 场时， 执行步骤 312; 当后向参考场对应块的运动矢量指向与偶场属于同一帧 相应的奇场时， 通过如下的公式， 计算得到后向运动矢量：
Figure imgf000010_0002
TDB是当前图片与前向参考图片在时域上的距离， TDD是前向参考图片与 后向参考图片在时域上的距离， MVF与 MVB分别是推导出来的相应 B帧当前块的 前向运动矢量和后向运动矢量； 结束步骤 31。
在步骤 41 中， 双向预测方法具体过程为： 对前向运动矢量和后向运动矢 量所指的两个预测参考块对应的像素取平均， 求得最终的双向预测参考块。 解码过程比较简单， 从码流中获得前向运动矢量后， 直接计算得到后向运 动矢量， 然后计算出双向预测参考块与块残差合并， 得出编码前的图像， 这完 全可以认为是编码过程的逆过程。 最后所应说明的是， 以上实施例仅用以说明本发明的技术方案而非限制， 尽管参照较佳实施例对本发明进行了详细说明， 本领域的普通技术人员应当理 解， 可以对本发明的技术方案进行修改或者等同替换， 而不脱离本发明技术方 案的精神和范围， 其均应涵盖在本发明的权利要求范围当中。

Claims

权 利 要 求
1、 一种用于视频编码的编码端双向预测方法， 其特征在于： 包括如下步 骤：
步骤 10、 对当前 B帧的每一图像块， 采用前向预测模式， 从前向参考图像 中获得当前图像块候选前向运动矢量；
步骤 20、 利用步骤 10得到的当前图像块的候选前向运动矢量， 计算得到 候选后向运动矢量， 获得进行双向预测所需的候选前向运动矢量和候选后向运 动矢量；
步骤 30、 利用步骤 20得到的当前图像块的候选前向运动矢量和候选后向 运动矢量， 通过双向预测方法得到候选双向预测参考块；
步骤 40、 在给定的搜索范围之内和 /或在匹配值小于或等于预先设定的匹 配阈值以前， 设定新的参考块， 重复步骤 10-30， 选取最优参考块；
步骤 50、将该图像块的由最优参考块确定的前向运动矢量、后向运动矢量 和块残差， 编入码流中。
2、 才艮据权利要求 1 所述的用于视频编码的编码端双向预测方法， 其特征 在于：所述步骤 10中，采用前向预测运动模式，获得候选前向运动矢量的过程， 包括如下步骤：
步骤 101、 如果是初次， 选定前向参考图片中与 B帧当前图像块位置相同 的图像块作为前向参考图片中设定的参考块， 执行步骤 102 ; 否则， 直接执行 步骤 102;
步骤 102、 前向参考图片中设定的参考块在前向参考图片中的位置与 B帧 当前图像块在当前图片中的位置作差运算， 所得到的向量即为候选前向运动矢 量； 结束步骤 10。
3、 根据权利要求 1 所述的用于视频编码的编码端双向预测方法， 其特征 在于： 所述的步骤 20中， 计算候选后向运动矢量的过程， 包括如下步骤： 步骤 110、 判断当前的图 «式， 如果为帧编码模式， 执行步骤 111; 如 果为场编码方式， 判断奇场还是偶场， 如果为奇场， 执行步骤 112; 如果为偶 场， 执行步骤 113;
步骤 111、 通过如下的公式， 计算得到候选后向运动矢量：
TD -TD
CMVB = -( ° B) X CMVF
B TDB F
TDB是当前图片与前向参考图片在时域上的距离， TDD是前向参考图片与 后向参考图片在时域上的距离，
Figure imgf000013_0001
CMVB分别是相应 B帧当前块的侯选前向与 侯选后向运动矢量； 结束步骤 20;
步骤 112、 通过如下的公式， 计算得到候选后向运动矢量：
TDDi-TDBi
CMVBJ = - ( D T'D Β'Ί χ MVFJ
TDB是当前图片与前向参考图片， 在时域上的距离， TDD是前向参考图片 与后向参考图片在时域上的距离， ^1^与 CMVB分别是推导出来的相应 B帧当前 块的前向与候选后向运动矢量； 结束步骤 20;
步骤 113、 当后向参考场对应块的运动矢量指向时域上早于当前场的某一 场时， 执行步骤 112; 当后向参考场对应块的运动矢量指向与偶场属于同一帧 相应的奇场时， 通过如下的公式， 计算得到候选后向运动矢量： TDn , + TDR ,
CMVB, = {-^- ~~ -) x C F,,,
丄
TDB是当前图片与前向参考图片在时域上的距离， TDD是前向参考图片与 后向参考图片在时域上的距离， 0¾^与 CMVB分别是推导出来的相应 B帧当前块 的候选前向运动矢量与候选后向运动矢量； 结束步骤 20。
4、 根据权利要求 1 所述的用于视频编码的双向预测方法， 其特征在于： 所述的步骤 30双向预测方法，具体过程为： 对候选前向运动矢量和候选后向运 动矢量所指的两个预测参考块对应的像素取平均，求得最终的双向预测参考块。
5、 根据权利要求 1 所述的用于视频编码的双向预测方法， 其特征在于： 所述的步骤 40中，搜索范围为以参考图片中与 B帧当前块位置相同的参考块为 中心的一定区域， 最大可遍及整个参考图片， 全部搜索范围中通过参考块计算 出的双向预测参考块与对应的 B帧当前块像素之差的绝对值之和最小的参考块 为最优参考块。
6、 根据权利要求 1 所述的用于视频编码的编码端双向预测方法， 其特征 在于：所述的步骤 40中匹配值为双向预测参考块与 B帧当前块对应像素之差的 绝对值的和， 匹配阈值为预先设定的匹配值， 如果匹配值小于或等于匹配阈值， 此时的参考块即为最优参考块。
7、 根据权利要求 1 所述的用于视频编码的编码端双向预测方法， 其特征 在于： 所述的块残差包括最优参考块与 B帧当前块之间对应像素的差异。
8、 一种用于视频编码的解码端双向预测方法， 其特征在于： 包括如下步 骤：
步骤 21、 从码流中解码， 获得前向运动矢量；
步骤 31、 利用步骤 21 中得到前向运动矢量时， 计算得到后向运动矢量， 得到了双向预测所需的前向运动矢量和后向运动矢量；
步骤 41、 利用步骤 31得到的当前图像块的前向运动矢量和后向运动矢量， 通过双向预测方法， 求得最终的双向预测参考块；
步骤 51、将步骤 41得到的预测参考块与码流中解码相应的得到的块残差合 并形成当前块图像块。
9、 根据权利要求 8 所述的用于视频编码的解码端双向预测方法， 其特征 在于： 所述的步骤 31中， 计算后向运动矢量的步骤包括：
步骤 310、 判断当前的图像模式， 如果为帧编码模式， 执行步骤 311 ; 如 果为场编码方式， 判断奇场还是偶场， 如果为奇场， 执行步骤 312; 如果为偶 场， 执行步骤 31 3;
步骤 311、 通过如下的公式， 计算得到后向运动矢量：
Figure imgf000015_0001
TDB是当前图片与前向参考图片在时域上的距离， TDD是前向参考图片与 后向参考图片在时域上的距离， MVF与 MVB分别是相应 B帧当前块的前向运动矢 量与后向运动矢量； 结束步骤 31 ;
步骤 312、 通过如下的公式， 计算得到后向运动矢量：
Figure imgf000016_0001
TDB是当前图片与前向参考图片， 在时域上的距离， TDD是前向参考图片 与后向参考图片在时域上的距离， MVF与 MVB分别是推导出来的相应 B帧当前块 的前向运动矢量与后向运动矢量； 结束步骤 31 ;
步骤 31 3、 当后向参考场对应块的运动矢量指向时域上早于当前场的某一 场时， 执行步骤 312 ; 当后向参考场对应块的运动矢量指向与偶场属于同一帧 相应的奇场时， 通过如下的公式， 计算得到后向运动矢量：
Figure imgf000016_0002
TDB是当前图片与前向参考图片在时域上的距离， TDD是前向参考图片与 后向参考图片在时域上的距离， MVF与 MVB分别是推导出来的相应 B帧当前块的 前向运动矢量和后向运动矢量； 结束步骤 31。
1 0、 根据权利要求 8所述的用于视频编码的解码端双向预测方法， 其特征 在于： 所述的步骤 41双向预测方法， 具体过程为： 对前向运动矢量和后向运动 矢量所指的两个预测参考块对应的像素取平均， 求得最终的双向预测参考块。
PCT/CN2004/000735 2003-09-12 2004-07-02 Bi-directional predicting method for video coding/decoding WO2005027520A1 (en)

Priority Applications (4)

Application Number Priority Date Filing Date Title
EP04738333.6A EP1672926B1 (en) 2003-09-12 2004-07-02 Bi-directional predicting method for video coding/decoding
BRPI0413945-3A BRPI0413945B1 (pt) 2003-09-12 2004-07-02 Método preditivo bidirecional para codificação e decodificação de vídeo
JP2006525606A JP4755095B2 (ja) 2003-09-12 2004-07-02 映像符号化の符号化側／復号化側に使用される双方向予測方法
US10/571,659 US8005144B2 (en) 2003-09-12 2004-07-02 Bi-directional predicting method for video coding/decoding

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CNB031570771A CN1225127C (zh) 2003-09-12 2003-09-12 一种用于视频编码的编码端/解码端双向预测方法
CN03157077.1 2003-09-12

Publications (1)

Publication Number Publication Date
WO2005027520A1 true WO2005027520A1 (en) 2005-03-24

Family

ID=34287139

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2004/000735 WO2005027520A1 (en) 2003-09-12 2004-07-02 Bi-directional predicting method for video coding/decoding

Country Status (7)

Country Link
US (1) US8005144B2 (zh)
EP (1) EP1672926B1 (zh)
JP (1) JP4755095B2 (zh)
KR (1) KR100897880B1 (zh)
CN (1) CN1225127C (zh)
BR (1) BRPI0413945B1 (zh)
WO (1) WO2005027520A1 (zh)

Cited By (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI551123B (zh) * 2011-11-07 2016-09-21 Ntt Docomo Inc A motion picture predictive coding apparatus, a motion picture prediction decoding apparatus, a motion picture prediction coding method, a motion picture prediction decoding method, a motion picture prediction coding program, and a motion picture prediction decoding program

Families Citing this family (68)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
BRPI0515723A (pt) * 2004-11-04 2008-08-05 Thomson Licensing método e aparelho para a decisão de modo rápido de b-quadros em um codificador de vìdeo
CN100384216C (zh) * 2004-12-31 2008-04-23 北京中星微电子有限公司 优化搜索算法的视频数据压缩方法及装置
CN100394398C (zh) * 2005-01-07 2008-06-11 深圳清华大学研究院 一种avs视频解码验证方法和装置
CN100405851C (zh) * 2005-11-18 2008-07-23 宁波中科集成电路设计中心有限公司 多模标准解码器中运动矢量预测复用设计方法
JP5165593B2 (ja) * 2006-02-02 2013-03-21 トムソン ライセンシング 組合せ参照双方向予測を用いて動き推定を行う方法および装置
US20070252895A1 (en) * 2006-04-26 2007-11-01 International Business Machines Corporation Apparatus for monitor, storage and back editing, retrieving of digitally stored surveillance images
JPWO2007136088A1 (ja) * 2006-05-24 2009-10-01 パナソニック株式会社 画像符号化装置、画像符号化方法、及び画像符号化用集積回路
WO2009035144A1 (ja) * 2007-09-12 2009-03-19 Sony Corporation 画像処理装置及び画像処理方法
CN101222646B (zh) * 2008-01-30 2010-06-02 上海广电（集团）有限公司中央研究院 一种适用于avs编码的帧内预测装置及预测方法
EP2557795A4 (en) * 2010-04-06 2015-07-08 Samsung Electronics Co Ltd METHOD AND APPARATUS FOR ENCODING A VIDEO AND METHOD AND APPARATUS FOR DECODING A VIDEO
WO2012008040A1 (ja) * 2010-07-15 2012-01-19 株式会社 東芝 画像符号化方法及び画像復号化方法
CN102006480B (zh) * 2010-11-29 2013-01-30 清华大学 基于视间预测的双目立体视频的编码及解码方法
KR101506446B1 (ko) * 2010-12-15 2015-04-08 에스케이 텔레콤주식회사 움직임정보 병합을 이용한 부호움직임정보생성/움직임정보복원 방법 및 장치와 그를 이용한 영상 부호화/복호화 방법 및 장치
GB2487200A (en) * 2011-01-12 2012-07-18 Canon Kk Video encoding and decoding with improved error resilience
EP2665271A4 (en) 2011-01-12 2014-06-18 Panasonic Corp VIDEO ENCODING METHOD AND VIDEO DECODING METHOD
US9066104B2 (en) 2011-01-14 2015-06-23 Google Inc. Spatial block merge mode
CN102710934B (zh) * 2011-01-22 2015-05-06 华为技术有限公司 一种运动预测或补偿方法
AU2012216026B2 (en) * 2011-02-10 2016-04-28 Sun Patent Trust Moving picture coding method, moving picture coding apparatus, moving picture decoding method, moving picture decoding apparatus, and moving picture coding and decoding apparatus
US10404998B2 (en) 2011-02-22 2019-09-03 Sun Patent Trust Moving picture coding method, moving picture coding apparatus, moving picture decoding method, and moving picture decoding apparatus
JP5358746B2 (ja) 2011-03-03 2013-12-04 パナソニック株式会社 動画像符号化方法、動画像符号化装置及びプログラム
PL3136727T3 (pl) 2011-04-12 2018-11-30 Sun Patent Trust Sposób kodowania ruchomych obrazów i urządzenie do kodowania ruchomych obrazów
MX2013012020A (es) 2011-05-24 2013-12-10 Panasonic Corp Metodo de codificacion de imagenes, aparato de codificacion de imagenes, metodo de decodificacion de imagenes, aparato de decodificacion de imagenes y aparato de codificacion /decodificacion de imagenes.
DK2717575T3 (en) 2011-05-27 2019-01-28 Sun Patent Trust PICTURE CODING PROCEDURE AND PICTURE CODES
US9485518B2 (en) 2011-05-27 2016-11-01 Sun Patent Trust Decoding method and apparatus with candidate motion vectors
US8989271B2 (en) 2011-05-31 2015-03-24 Panasonic Intellectual Property Corporation Of America Decoding method and apparatus with candidate motion vectors
SG194746A1 (en) * 2011-05-31 2013-12-30 Kaba Gmbh Image encoding method, image encoding device, image decoding method, image decoding device, and image encoding/decoding device
GB2491589B (en) 2011-06-06 2015-12-16 Canon Kk Method and device for encoding a sequence of images and method and device for decoding a sequence of image
CA2836063C (en) 2011-06-30 2020-06-16 Panasonic Corporation Image encoding and decoding method and device for generating predictor sets in high-efficiency video coding
KR20140034292A (ko) * 2011-07-01 2014-03-19 모토로라 모빌리티 엘엘씨 움직임 벡터 예측 설계 간소화
CN103718558B (zh) 2011-08-03 2017-04-19 太阳专利托管公司 运动图像编码方法及装置、解码方法及装置和编解码装置
MX343471B (es) * 2011-08-29 2016-11-07 Ibex Pt Holdings Co Ltd Metodo para generar un bloque de prediccion en modo de prediccion de vector de movimiento avanzada (amvp).
EP2770738A4 (en) 2011-10-19 2016-07-27 Panasonic Ip Corp America IMAGE ENCRYPTION METHOD, IMAGE ENCRYPTION DEVICE, IMAGE KEYING METHOD AND IMAGE WRITING DEVICE
BR112014010539A2 (pt) 2011-11-04 2017-04-18 Motorola Mobility Llc escalonamento de vetor de movimento para rede de vetor de movimento não uniforme
US9531990B1 (en) * 2012-01-21 2016-12-27 Google Inc. Compound prediction using multiple sources or prediction modes
US8908767B1 (en) 2012-02-09 2014-12-09 Google Inc. Temporal motion vector prediction
US8947449B1 (en) 2012-02-21 2015-02-03 Google Inc. Color space conversion between semi-planar YUV and planar YUV formats
US8737824B1 (en) 2012-03-09 2014-05-27 Google Inc. Adaptively encoding a media stream with compound prediction
WO2013136678A1 (ja) * 2012-03-16 2013-09-19 パナソニック株式会社 画像復号装置および画像復号方法
US9172970B1 (en) 2012-05-29 2015-10-27 Google Inc. Inter frame candidate selection for a video encoder
US11317101B2 (en) 2012-06-12 2022-04-26 Google Inc. Inter frame candidate selection for a video encoder
KR101932916B1 (ko) 2012-06-28 2018-12-27 삼성전자 주식회사 모션 추정 시스템 및 그 방법, 디스플레이 컨트롤러, 및 전자 장치
US9185414B1 (en) 2012-06-29 2015-11-10 Google Inc. Video encoding using variance
US9503746B2 (en) 2012-10-08 2016-11-22 Google Inc. Determine reference motion vectors
US9485515B2 (en) 2013-08-23 2016-11-01 Google Inc. Video coding using reference motion vectors
US9628790B1 (en) 2013-01-03 2017-04-18 Google Inc. Adaptive composite intra prediction for image and video compression
US9374578B1 (en) 2013-05-23 2016-06-21 Google Inc. Video coding using combined inter and intra predictors
US9313493B1 (en) 2013-06-27 2016-04-12 Google Inc. Advanced motion estimation
CN104427345B (zh) * 2013-09-11 2019-01-08 华为技术有限公司 运动矢量的获取方法、获取装置、视频编解码器及其方法
US9609343B1 (en) 2013-12-20 2017-03-28 Google Inc. Video coding using compound prediction
US9438910B1 (en) 2014-03-11 2016-09-06 Google Inc. Affine motion prediction in video coding
WO2016008161A1 (en) * 2014-07-18 2016-01-21 Mediatek Singapore Pte. Ltd. Temporal derived bi-directional motion vector predictor
CN104363461B (zh) * 2014-11-04 2017-12-22 中国科学技术大学 视频帧的错误隐藏方法及应用其的视频解码方法
CN106034236B (zh) 2015-03-19 2019-07-19 阿里巴巴集团控股有限公司 一种hevc编码最佳参考帧的选择方法、装置及编码器
CN104717513B (zh) * 2015-03-31 2018-02-09 北京奇艺世纪科技有限公司 一种双向帧间预测方法及装置
CN107787582A (zh) * 2015-06-10 2018-03-09 三星电子株式会社 使用用于自适应加权预测的语法信令对图像进行编码或解码的方法和设备
CN109996080B (zh) * 2017-12-31 2023-01-06 华为技术有限公司 图像的预测方法、装置及编解码器
CN110324637B (zh) 2018-03-29 2023-11-17 华为技术有限公司 一种双向帧间预测方法及装置
KR20230143620A (ko) * 2018-07-06 2023-10-12 미쓰비시덴키 가부시키가이샤 동화상 복호 방법, 동화상 복호 장치, 동화상 부호화 방법 및 동화상 부호화 장치
CN109145744B (zh) * 2018-07-19 2020-08-18 西安理工大学 一种基于自适应预测模式的lstm网络行人重识别方法
US11051034B2 (en) * 2018-10-08 2021-06-29 Qualcomm Incorporated History-based motion vector predictor
CN109525843B (zh) * 2018-10-23 2022-06-10 深圳市网心科技有限公司 skip块判断方法、电子装置及计算机可读存储介质
WO2020138958A1 (ko) 2018-12-27 2020-07-02 에스케이텔레콤 주식회사 양방향 예측 방법 및 영상 복호화 장치
KR20200081201A (ko) * 2018-12-27 2020-07-07 에스케이텔레콤 주식회사 양방향 예측 방법 및 영상 복호화 장치
CN116996683A (zh) * 2019-01-03 2023-11-03 北京大学 视频图像处理方法与装置
CN112954339B (zh) * 2019-03-11 2022-08-26 杭州海康威视数字技术股份有限公司 一种编解码方法、装置及其设备
CN113709486B (zh) * 2019-09-06 2022-12-23 杭州海康威视数字技术股份有限公司 一种编解码方法、装置及其设备
CN111754429B (zh) * 2020-06-16 2024-06-11 Oppo广东移动通信有限公司 运动矢量后处理方法和装置、电子设备及存储介质
CN115633216B (zh) * 2022-09-05 2024-05-28 北京智源人工智能研究院 时域运动一致性视频生成模型的训练方法和视频生成方法

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1995021503A1 (en) * 1994-02-02 1995-08-10 British Broadcasting Corporation Video motion estimation
EP0863674A2 (en) 1997-03-07 1998-09-09 General Instrument Corporation Prediction and coding of bi-directionally predicted video object planes for interlaced digital video
CN1272288A (zh) * 1998-05-20 2000-11-01 皇家菲利浦电子有限公司 编码作为图象序列组织的信息流
CN1302510A (zh) * 1999-04-30 2001-07-04 皇家菲利浦电子有限公司 选择b帧编码模式的视频编码方法
WO2003041382A2 (en) 2001-11-07 2003-05-15 Motorola Inc Scalable video transmissions

Family Cites Families (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5198901A (en) * 1991-09-23 1993-03-30 Matsushita Electric Corporation Of America Derivation and use of motion vectors in a differential pulse code modulation system
US5293229A (en) * 1992-03-27 1994-03-08 Matsushita Electric Corporation Of America Apparatus and method for processing groups of fields in a video data compression system
JPH07162864A (ja) * 1993-12-08 1995-06-23 Ricoh Co Ltd 動きベクトル検出方法
KR0171145B1 (ko) * 1995-03-20 1999-03-20 배순훈 이미지 부호화 시스템용 움직임 보상 장치
KR100209793B1 (ko) * 1995-10-28 1999-07-15 전주범 특징점 기반 움직임 추정을 이용하여 비디오 신호를 부호화 및 복호화하는 장치
JP3263807B2 (ja) * 1996-09-09 2002-03-11 ソニー株式会社 画像符号化装置および画像符号化方法
JP2001045492A (ja) * 1999-08-03 2001-02-16 Canon Inc 動画像処理方法
JP3674535B2 (ja) * 2001-05-08 2005-07-20 日本電気株式会社 動画像符号化方法および装置
TR201802383T4 (tr) * 2001-11-06 2018-03-21 Panasonic Ip Corp America Hareketli resim kodlama usulü ve hareketli resim kod çözme usulü.

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO1995021503A1 (en) * 1994-02-02 1995-08-10 British Broadcasting Corporation Video motion estimation
EP0863674A2 (en) 1997-03-07 1998-09-09 General Instrument Corporation Prediction and coding of bi-directionally predicted video object planes for interlaced digital video
CN1272288A (zh) * 1998-05-20 2000-11-01 皇家菲利浦电子有限公司 编码作为图象序列组织的信息流
CN1302510A (zh) * 1999-04-30 2001-07-04 皇家菲利浦电子有限公司 选择b帧编码模式的视频编码方法
WO2003041382A2 (en) 2001-11-07 2003-05-15 Motorola Inc Scalable video transmissions

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
GARY SULLIVAN ET AL.: "Technical Description of ITU-T VCEG Draft H. 26L Algorithm in Response to Video and DCinema CfPs", ITU STUDY GROUP 16 - VIDEO CODING EXPERTS GROUP -ISO/IEC MPEG & ITU-T VCEG
See also references of EP1672926A4 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
TWI551123B (zh) * 2011-11-07 2016-09-21 Ntt Docomo Inc A motion picture predictive coding apparatus, a motion picture prediction decoding apparatus, a motion picture prediction coding method, a motion picture prediction decoding method, a motion picture prediction coding program, and a motion picture prediction decoding program
TWI597974B (zh) * 2011-11-07 2017-09-01 Ntt Docomo Inc Dynamic image prediction decoding method and dynamic image prediction decoding device

Also Published As

Publication number Publication date
US8005144B2 (en) 2011-08-23
EP1672926B1 (en) 2013-05-22
BRPI0413945A8 (pt) 2018-04-17
EP1672926A4 (en) 2011-08-03
US20070110156A1 (en) 2007-05-17
JP4755095B2 (ja) 2011-08-24
BRPI0413945A (pt) 2006-10-31
KR100897880B1 (ko) 2009-05-18
JP2007505529A (ja) 2007-03-08
KR20070026317A (ko) 2007-03-08
CN1225127C (zh) 2005-10-26
CN1525762A (zh) 2004-09-01
EP1672926A1 (en) 2006-06-21
BRPI0413945B1 (pt) 2019-05-14

Similar Documents

Publication Publication Date Title
WO2005027520A1 (en) Bi-directional predicting method for video coding/decoding
TWI392374B (zh) 於可調節視訊編碼中用以使用框率上升轉換技術之方法與裝置
EP2002662B1 (en) Apparatus and method of enhanced frame interpolation in video compression
US8073048B2 (en) Method and apparatus for minimizing number of reference pictures used for inter-coding
JP5081305B2 (ja) フレーム間予測符号化の方法および装置
TW200942045A (en) Method for video coding
WO2006103844A1 (ja) 符号化装置と方法、ならびに復号装置と方法
US20070133689A1 (en) Low-cost motion estimation apparatus and method thereof
JP2009260421A (ja) 動画像処理システム、符号化装置、符号化方法、符号化プログラム、復号化装置、復号化方法および復号化プログラム
WO2005067298A1 (en) The method of determining the reference block in direct coding mode
Zhang et al. Distributed soft video broadcast with variable block size motion estimation
JP4642033B2 (ja) 参照フレームの数を固定する符号化方式で画像の参照ブロックを取得する方法
Shaikh et al. Video compression algorithm using motion compensation technique
Lee et al. Low bit rates video coding using hybrid frame resolutions
Goh et al. Real-time software MPEG-2 TO H. 264 video transcoding
KR100955828B1 (ko) 참조 프레임 선택을 이용한 고성능 멀티미디어 코덱에 사용되는 움직임 추정 연산 방법
Ji et al. An Avs-P2 encoding scheme based on motion compensation de-interlacing information
Kumar et al. Extraction of Desired Partial Block of a Frame from MPEG Video Stream
JP2002374536A (ja) 符号化装置
Jiang et al. New multiple description scalable video coding based on redundant wavelet
Lai et al. A quantized transform-domain motion estimation technique for H. 264 secondary SP-Frames
KR20040091939A (ko) 인트라 휘도 예측 모드 결정방법 및 장치
Jeoti et al. Introducing SKIP mode in transform domain Distributed Video Coding
WO2005036465A1 (en) Video encoding method and device
Furht Jose Luis Martinez 2 and Gerardo Fernández Escribano 2

Legal Events

Date Code Title Description
AK Designated states

Kind code of ref document: A1

Designated state(s): AE AG AL AM AT AU AZ BA BB BG BW BY BZ CA CH CN CO CR CU CZ DK DM DZ EC EE EG ES FI GB GD GE GM HR HU ID IL IN IS JP KE KG KP KZ LC LK LR LS LT LU LV MA MD MK MN MW MX MZ NA NI NO NZ PG PH PL PT RO RU SC SD SE SG SK SY TJ TM TN TR TT TZ UA UG US UZ VN YU ZA ZM

AL Designated countries for regional patents

Kind code of ref document: A1

Designated state(s): GM KE LS MW MZ NA SD SZ TZ UG ZM ZW AM AZ BY KG MD RU TJ TM AT BE BG CH CY DE DK EE ES FI FR GB GR HU IE IT MC NL PL PT RO SE SI SK TR BF CF CG CI CM GA GN GQ GW ML MR SN TD TG

121 Ep: the epo has been informed by wipo that ep was designated in this application
WWE Wipo information: entry into national phase

Ref document number: 2006525606

Country of ref document: JP

WWE Wipo information: entry into national phase

Ref document number: 2004738333

Country of ref document: EP

WWE Wipo information: entry into national phase

Ref document number: 1020067007029

Country of ref document: KR

WWP Wipo information: published in national office

Ref document number: 2004738333

Country of ref document: EP

ENP Entry into the national phase

Ref document number: PI0413945

Country of ref document: BR

WWE Wipo information: entry into national phase

Ref document number: 2007110156

Country of ref document: US

Ref document number: 10571659

Country of ref document: US

WWP Wipo information: published in national office

Ref document number: 10571659

Country of ref document: US