A Predictive VQ Based Video Compression Scheme

Yue-Yang ZHANG,LI Zhen,Hua TUO,LU Zhe-Ming
2007-01-01
Abstract:As an efficient technology for lossy data compression, vector quantization (VQ) is widely used in the field of image coding with the virtues of simple algorithm and high compression rate. ITU-H.264 is a video coding standard published in March, 2005. The framework of H.264 is not much different compared to those previous standards, but it improves each function by adopting mature techniques, so H.264 has the highest coding efficiency, yet with highest complexity. Generally speaking, H.264 is still a video compression based on transformation. Considering the weakness of high complexity of H.264 and the advantage of simple structure and low coding complexity of VQ technology, this paper proposes a novel predictive VQ based video compression scheme, which uses H.264 standards coding method for reference and combines VQ techniques. It uses the minimax partial distortion competitive learning method to generate a public codebook, making sure that only codeword indices need to be transmitted after video sequence compression. It uses the mean-distance-ordered partial codebook search algorithm for fast code vector searching, which uses the mean of input vectors to reduce the computing expense dramatically without performance loss. Experiment results show that this scheme, with the reduction of the computing complexity and the promotion of coding efficiency, gives a tradeoff of expense and quality. It can be used in real time occasions and mobile handset equipments with limited computing capability.
What problem does this paper attempt to address?