Common video compression standards

MPEG-X

MPEG-1

MPEG-1 is the first video coding standard developed by ISO/IEC MPEG working group. Specifically for CD media customization, completed in 1991, 1992 was officially approved as the world standard. For motion compensation, MPEG-1 uses block-based motion compensation and refines the motion vector accuracy to half a pixel, which allows bidirectional prediction between frames. In addition, MPEG-1 also employs adaptive quantization with a separate quantization scaling factor for each frequency band in order to facilitate the improvement of people’s subjective visual experience. MPEG-1 defines the coding structure of three image types I, P and B in the prediction coding module for the first time, that is, I frame is intra-frame prediction coding, P frame is forward prediction coding and B frame is bidirectional prediction coding. All macroblocks in I frame use intra-frame coding method, and all macroblocks in P frame and B frame can use either intra-frame coding method or inter-frame coding method.

MPEG-2

MPEG-2 video coding standard is usually considered an ISO standard, but it is an official project developed jointly by ISO/IEC JTC1 and ITU-T organization, which was finalized in 1994. It is mainly used in Standard Definition TV (SDTV), which is their mainstream coding standard. MPEG-2 is still used today in Digital Video Broadcasting (DVB) and high definition television. MPEG-2 inherits many characteristics of H.261 and MPEG-1, and makes improvements in system and transmission, adds many provisions and further improves the low bit rate compression and multi-channel expansion based on MPEG-1. In addition, MPEG-2 supports both interlaced scanning and progressive scanning. The coding framework it uses is still the mainstream today.

MPEG-4

MPEG-4 uses the algorithm of the video coding standard H.263 as a starting point, so that the MPEG-4 codec can be compatible with any video image bit stream encoded by H.263, but at the same time MPEG-4 develops some additional features that can improve the compression efficiency. 8ⅹ8DCT transform and scalar quantization are adopted in MPEG-4, but it is compatible with H.263 standard, so it supports two kinds of scalar quantization methods, i. e., H.263 style and MPEG style.

Unlike the previous two standards in the same series, which focus solely on coding efficiency, MPEG-4 optimizes data compression at the object level and uses video coding methods that are closely related to the video content. The new techniques used include quarter-pixel motion compensation, motion vector field adaptive search, feature-based fast and robust global motion estimation, etc. MPEG-4 is widely used in video phone, wireless communication, video E-mail and so on.

H.26x

H.261

H.261 is the first mainstream video coding standard, which was originally developed to enable Integrated Services Digital Network (ISDN) networks with 64kbps of bandwidth. ISDN is used to transmit video signals and ensure acceptable subjective quality of video. It is mainly used for video with bit rate between about 40kbps and 2Mbps. H.261 adopts a hybrid coding framework based on “prediction/transform”. Time has proved that it has achieved great success, so that the subsequent video coding standards are based on its original framework to continuously improve the quality of compressed video images and reduce the number of bits required for data transmission. The main techniques include inter-frame prediction based on motion estimation and motion compensation, discrete cosine transform, quantization and entropy coding. As the first mainstream standard, it is a pity that H.261 can only adopt intra coding and forward prediction coding, and bidirectional prediction technology has not been used in the coding framework of H.261.

H.262

H.262 is the first video coding standard developed by VCEG and MPEG, so it is also called MPEG-2. The final version and the first revision were completed in 1994 and 2000 respectively. Compared with other standards, H.262 has the widest application range and the highest application frequency in audio and video products such as TV sets and cameras.

H.263

H.263 is a new standard developed after the success of H.261. It uses basic source coding methods similar to MPEG-2, including intra-frame, inter-frame and skip coding modes. However, its main techniques, such as median motion vector prediction and VLC coding, have made significant progress to achieve higher video quality at lower bit rates. The first version of H.263 uses eight coding tools (Annexes A to G), which can effectively improve the compression efficiency. Where annexes D,E,F, and G control the source coding options to improve the compression performance. Annex D provides for the processing of points other than the reference picture of the motion vector and the use of motion estimation vectors longer than the H.263 base file. Annex F specifies the use of overlapping block motion compensation as well as the allocation of four motion vectors to each macroblock, each of which is divided into 8ⅹ8 sub-blocks, that is, variable block size is used. Therefore, 8ⅹ8 coding modes are added to the set of possible macroblock modes.

H.264

H.264/AVC is the last Joint project of the Joint Video Team (JVT), which is jointly organized by ITU-T’s VCEG and ISO/IEC’s MPEG. It is also the most successful video coding standard. H.264/AVC design includes video coding layer and network adaptation layer. Many new techniques are applied in H.264, such as multi-reference frame motion compensation, which increases the number of reference frames from 1 or 2 to 32. In the process of motion estimation and motion compensation, the block size can be selected from 16×16 to 4×4 to achieve finer division of moving objects. In addition, the pixel accuracy of motion compensation is refined to 1/4. The predicted value of the luminance component can reach 1/2 pixel, which is realized by a six-tap filter to reduce the ringing effect. H.264/AVC has higher compression efficiency than other standards, and the unified data format can be quickly transmitted over wireless channels, so it is widely used in the field of digital video compression and transmission. At present, it is mainly used in digital TV, satellite TV, video conference, remote communication, remote medical treatment, distance education and other fields.

H.265

H.265 is a Video Coding standard started by the JCT-VC research group established by ITU-T and ISO/IEC in 2010, named as High Efficiency Video Coding (HEVC). With the increasing variety of Internet services, the increasing popularity of high-definition video, and the emergence of ultra-high definition video (e.g., 4k×2k or 8k×4k resolution), the coding efficiency of H.264/MPEG becomes limited. In addition, the traffic caused by video applications on mobile devices and tablets, as well as the transmission requirements of video-on-demand services, pose severe challenges to today’s networks. HEVC is designed to address all existing applications that H.264/MPEG cannot, and the main goal is to substantially improve the compression performance of the existing standard, reducing the consumed bit rate by 50% without degrading the video quality. HEVC adopts a Coding Tree Unit (CTU), which is similar to the coding unit in previous coding standards. Different from H.264, HEVC supports tree structure and quadtree-like signaling to divide CTU into smaller blocks, and distinguishes between prediction unit and transform unit. In addition, more intra prediction modes are added to HEVC, and two different modes are introduced at the inter prediction level to allow motion vectors to be directly obtained from neighboring coding units, as well as optimized deblocking and sampling point adaptive offset filters.

AVS

AVS standard is a source coding standard with independent intellectual property rights of China. It is a series of standards set up according to international open rules according to the demand of China’s audio and video industry and led by Chinese research institutions and enterprises in the field of digital audio and video with extensive participation of relevant international institutions and enterprises. Two generations of AVS standards have been developed.

AVS1

In 2002, the AVS Working Group began to draft the first generation AVS standard. AVS1 includes four main technical standards of system, video, audio, digital rights management and other supporting standards of conformance testing. The first generation of AVS standards includes national standards AVS1 and AVS+. The compression efficiency of AVS+ is as high as that of H.264/AVC. There are already thousands of AVS+ HD content on the star broadcast.

AVS2

In order to better apply to ultra-high definition TV programs and support the digital media industry in the next few years, and strive to play a key role in the formulation of relevant international standards, the working group has carried out the research of the second generation standard AVS2. The primary application target is ultra-high definition video, which supports efficient compression of ultra-high resolution (above 4K) and high dynamic range video. The compression efficiency of AVS2 is more than twice that of the previous generation standards AVS+ and H.264/AVC, and more than that of the international standard HEVC/H.265. AVS2 also supports efficient coding of 3D video, multi-view video and virtual reality video. Efficient lossy and lossless coding of stereo and multichannel audio; Efficient coding of surveillance video; New media service for Triple play.

AVS3

The proposal and improvement of AVS3 has laid a solid foundation for the localization of 8K ultra-HD TV transmission, distribution and terminal presentation system equipment in China. The successful launch of 8K ultra-HD test channel of China Media Group is a concentrated display of scientific and technological innovation in the field of ultra-HD in China.

Your task

Tell me about your knowledge of the AVS3 coding standard.

Share your thoughts and ideas in the comments below.