As the development of multi-media technology and the improvement ofhardware performance, the application of digital video in China becomeswidely used. As its base technology, the digital video compression technologybecomes more important. The most advanced technology standard in videocompression field is the H.264 of ITU-T (i.e. ISO-MPEG4 Part10/AVC), itshigh performance in compression has been proved in many application, but itsexpensive license limits its use in China. How to develop video compressiontechnology which we have intelligent property right and compatible withH.264 becomes very pressure and important.This dissertation is based on mentor's Optimal Haar Wavelet Theory, fathermore researched multi-band haar wavelet's property, fast algorithm andapplication in video codec. The main works are as follows:(1) Based the new features that the application of 4x4 transformation invideo codec becomes more and more active, we mainly researched the fastalgorithm of 4-band Optimal Haar Wavelet, and implemented it in software.(2) Aim at shilling the disadvantage of mix use of multi transforms' inH.264, based the idea of multi-resolution analysis, use 4-band haar wavelet todecompose images, use motion algorithm and encode technology like H.264,we proposed a new set of video codec algorithm, and implemented algorithmin software for 4-band Optimal Haar Wavelet and DCT(it is also 4-band haarwavelet, but not optimal).(3) Trained corresponding quantize matrix for 4-band Optimal HaarWavelet Transformation.(4) Tested and compared the performance of the new codec algorithmwith H.264.The results show that in new algorithm, 4-band Optimal HaarWavelet compared with 4x4 DCT, the performance of the former is not worsethan the latter, part of results is better than DCT. The results also show, the compression efficiency of the new codec algorithm based on 4-band OptimalHaar Wavelet is closed to the one of H.264.When below 300kbps, the betterone is H.264, when exceeded the point, the winner is the new algorithm. |