| Firstly, this article introduces the encoding technology used by H.264/AVC which is anew generation of video coding standard, it has been improved and upgraded from theprevious video coding standard. Because of the new encoding technology, the video dataencoded by H.264/AVC could be further compressed, and the performance is excellent inreal-time video communication and video storage, which makes it suitable for interactive ornon-interactive applications. Compared with H.263and MPEG-4encoding standard, the bitrate could be reduced to fifty percent, and the saving of the bandwidth is considerable.Secondly, this article analyzes several key techniques used in the H.264/AVC codingstandard. The technique includes the application of new prediction technology (theintra-frame/inter-frame) and13kinds of prediction mode in the frame within the codingalgorithm and not constant size block motion compensation in inter-frame prediction mode.The use of the integer DCT and Hadamard transform is also explained. Next, the quantizationis talked about, the decoded image resolution is different based on the multiple levels ofquantitative choice, the encoding bits are of course different, selecting different quantitativelevels is in order to accommodate the different network environment and applications. Andthen the deblocking filtering technology is discussed. When the prediction is beginning, theencoded block will have to be used to reconstruct itself, but the block effect will exist if thereconstructed block is not filtered, the block effect is similar to mosaic, in order to removethis effect, the deblocking filtering is needed, it is also an important technique to obtain thehigh-definition image. The entropy coding techniques are explored, which include CAVLCand CABAC,the two kinds of entropy coding techniques are based on context andself-adaptive.However, the improvement of compression ratio for H.264/AVC is companied with highcomplexity, so the purpose of this paper is to reduce the complexity of algorithm, is toimprove coding efficiency at the same time. The innovation of this article is put forward twosimple and efficient fast frame mode selection algorithms.The paper proposes a mode selection algorithm between intra44and intra1616, thisalgorithm takes full advantage of the similarity between a pixel and the pixel around withinthe current macroblock, and when the similarity is very large, it illustrates the current area isflat, and when the situation is contrary, it means the area is not flat. According to doublethreshold values to judge whether the current macroblock is in flat area, if those characteristics is not obvious, the original search algorithm will be also used to select the bestprediction mode.This paper also presents an optimization method with which the nine kinds of intra4×4block can be reduced to four. The algorithm would make full use of the texture direction ofimage block, the pixels of a4×4block are similar in a certain direction, through thecomparison of the consistent degree between4main directions and the actual4×4blocktexture direction, to select the corresponding mode according to the most possible direction,and then combined with its adjacent mode and DC mode as the final candidate modes. If thefast method is used only, the experimental verification shows that the encoding time can besaved about twenty percent and the PSNR changes little.The above two kinds of algorithm can be combined with each other and the coding timecan be further shorten. The result of experiments shows that compared with full searchalgorithm, the fast mode selection algorithm can reduce the coding time by about thirtypercent,the coding time is greatly reduced, the coding efficiency is improved, the as a result,the real-time visual communication could be further enhanced. Compared with other fastmode selection algorithm, the algorithm of this paper is very simple, and the performance isvery well. |
PDF Full Text Request