Research Of Lossless Video Coding Algorithm Based On Prediction

Digital video is one of the important mediums of access to information, however，the huge storage volume of digital video has brought great challenges to the computerstorage capacity and channel bandwidth，which seriously hampers the development ofvideo techniques. To solve this problem, video coding techniques have become the keymeans. The storage capacity of encoded video is greatly reduced, so that the cost ofstorage and transmission can also be largely reduced, which lays the foundation of itsapplication on the Internet.Video coding techniques have made great progress, but most of the studies are lossyvideo coding, however there are few excellent lossless coding methods. Meanwhile,many fields urgently need efficient lossless coding methods. Due to this status quo, thisthesis uses "lossless video coding" as research topics and proposes a lossless videocoding algorithm based on prediction.Based on the theory of fuzzy logic, this thesis introduces the concept of similarityfor the description of similar degree of the pixels. Intra-frame similarity is used todescribe the internal smoothing degree of the image block and determine the image’ssub-block mode—8×8pattern and4×4pattern. Interframe similarity is used to describethe matching degree of target block and predicted block, which can be the matchingcriteria of motion estimation. The comparison of the values of intra-frame similarity andinterframe similarity is used to determine the coding mode—intra-frame predictionbased on GAP predictor, interframe prediction based on DS search algorithm,intra-interframe mixed prediction. The subblock image is encoded according to thesub-block mode and coding mode, then the context-based Golomb-Rice coding methodis used to code the prediction residual. In order to test the prediction performance of theproposed algorithm, simulation experiments in Matlab platform are carried out. Theproposed algorithm is compared with MED prediction, GAP prediction and interframeprediction. Simulation results show that the proposed algorithm has better performancein prediction, which proves that the proposed algorithm has strong feasibility. A losslessvideo coding system based on the proposed algorithm is completed. A set of standardvideo sequences are used to test the compression ratio of the system. By comparisonwith the existing lossless coding algorithms based on prediction, the compression ratiohas certainly improved, which is the expected experimental results. Based on the proposed algorithm and object-oriented method, the lossless videocoding system is completed in VC++6.0development environment, which is capable ofloading, playing, encoding and decoding videos of YUV file format. Each module of thesystem core algorithm is independent, which can be used for secondary development.The application of lossless coding system proves that the proposed algorithm has strongfeasibility and has achieved the goal of the design of more efficient compressionalgorithm.