|
Communication and Information, Nanjing University of Posts and Telecommunications Institute of Xiao-Fei Li 1, AVS Standard in AVS is China's independent development of audio and video coding standard. AVS Working Group was established in June 2002, in just over a year time, considered 182 proposals, 41 have adopted the proposal, AVS video part of the final version in December 2003. AVS Standard in order to present the world's most advanced MPEG-4 AVC/H.264 framework as a starting point for extensive independent development of Chinese standards for digital video applications, emphasizing intellectual property, taking full account of the difficulty to achieve. AVS and MPEG-4 AVC/H.264 standard of the main technical differences include: 8x8 integer transform, quantization, intra prediction, 1 / 4 pixel precision interpolation, a special inter prediction, motion compensation, entropy coding, deblocking filtering ring. AVS 8x8 integer transform used, it can be 16-bit processors to achieve without mismatch, to overcome the MPEG-4AVC / H.264 video compression coding before all the international standards adopted by the existence 8x8DCT transform the inherent mismatch problem. The MPEG-4 AVC/H.264 4x4 integer transform used in the high-resolution video images on to the relevant performance is less than 8x8 transformation effective. AVS uses 64 quantitative, can adapt to different applications and services on the rate and quality requirements. AVS used in the current 8x8 transform and quantization scheme greatly reduces the difficulty of realizing the chip (you can greatly reduce the chip power consumption for mobile products that are very valuable). Intra prediction with reference MPEG-4AVC/H.264 ideas, AVS uses intra prediction technology, with adjacent blocks of pixels predicts the current block, while using the direction on behalf of a variety of spatial texture prediction model. However, AVS luma and chroma 8x8 block intra prediction is to the unit. Luminance prediction models using 5, chroma block with four kinds of prediction models, and that there are four kinds of modes 3 and luminance the same as the forecast model. In the very premise of encoding quality, AVS uses fewer prediction model, making it more concise, much less implementation complexity. Inter-frame motion compensation coding is a hybrid coding framework for one of the most important part. AVS standard with a 16 × 16,16 × 8,8 × 16 and 8 × 8 block mode motion compensation, in addition to MPEG-4AVC/H.264 away the standard of 8 × 4,4 × 8,4 × 4 The block mode the purpose is to better describe the movement of objects, to improve the accuracy of the motion search. Experiments show that for high-resolution video, AVS can use the block model has enough precision to express movement of objects. Less of the block mode, can reduce the motion vector and block mode of transmission costs, thereby enhancing the compression efficiency, reduce Decoding complexity. AVS and MPEG-4 AVC / H.264 are used 1 / 4 pixel accuracy of motion compensation. MPEG-4AVC/H.264 a 6-tap filter for half pixel interpolation and the bilinear filter is 1 / 4 pixel interpolation. The AVS uses a different 4-tap filter for half-pixel interpolation, and 1 / 4 pixel interpolation, in the case without compromising the performance needed to reduce the interpolation reference pixels, reducing the data access bandwidth requirements, which in high resolution rate video compression applications, it is very meaningful. In the traditional video coding standard (MPEG-x series and H.26x series), the bi-directional predicted frames B frames have only a forward reference frame and a backward reference frame, while the forward prediction frame P frame only a forward reference frame. The recent MPEG-4 AVC/H.264 full use of image correlation between the time domain, allowing P frames and B frames with multiple reference frames, can have up to 31 reference frames. Multi-frame information technology to increase the compression efficiency will also greatly increase the storage space and data access costs. AVS in P frames can use up to two of the former to the reference frame, B frames before and after use of a reference frame, P frames with B frames (including backward reference frame) the same reference frame, the reference frame storage space and data no less than the cost of access to traditional video coding standard large, and precisely the full use of the resources must be reserved. AVS in the B frame prediction using two-way direct mode (directmode), symmetric mode (symmetric mode) and the jump mode (skipmode). Using the symmetric mode, the bit stream before transmission only to motion vectors, motion vector after the forward motion vector can be derived, after saving the overhead of encoding the motion vector. For direct mode, the front block of forward and backward motion vectors are from the reference image corresponding to the position after the block motion vector derived, without transmission of motion vector, motion vector and therefore can save the encoding overhead. Skip mode motion vectors derived in the same way and direct mode, skip mode coding block motion compensation residuals are all zero, that is, the mode macro block transfer mode signal only, without the need to transmit motion vectors, residual compensation and other additional information. AVS entropy coding using adaptive variable length coding techniques. Entropy coding in AVS, all the syntax elements and residual data are based index map of Columbus in the form of binary code bit stream. Columbus code exponential advantages: on the one hand, its relatively low hardware complexity, closed formula can resolve code word, no look-up table; the other hand, elements of which can encode the flexibility to determine the probability distribution with k-index Columbus coding, if the k selected and properly, you can approximate entropy coding efficiency. Of the prediction residual transform coefficient block by scanning the formation of (level, run) on the string, level, run is not an isolated incident and there is a strong correlation, in the AVS in the level, run jointly by two-dimensional code and, based on the current level, run the probability distribution of the different trends, adaptive to change the order of Exponential-Golomb coding. AVS-Video currently defines a notch (profile) that benchmark grade. The benchmark grade are divided into four levels (level), corresponding to high-definition and standard definition applications. And MPEG-4AVC / H.264's baselineprofile compared, AVS-Video added B frame, interlace and other technology, so the compression efficiency increases with the MPEG-4AVC / H.264 than the main profile, but also reduced CABAC and so difficult to achieve the technology. AVS standard application of the main characteristics of targeted, technology and targeted. Therefore, high-resolution applications, the compression efficiency significantly higher than now in digital TV, optical storage media commonly used in MPEG-2 video to the next level. In the very premise of compression efficiency, but also higher than MPEG-4AVC / H.264 main profile of the implementation of complex The e-commerce company in China offers quality products such as China gravity die casting , China aluminum gravity casting, and more. For more , please visit green sand casting today!
Related Articles -
China gravity die casting, China aluminum gravity casting,
|
