Research On Scalable Video Encoding/Decoding System Based On Db2 Wavelet

With the development of computer and communication technique, the requirement for video information becomes more and more diverse. Conventional video coding can not provide the bit-stream that can be adapted to different requirements and transferring conditions, it is not suitable for the current network transferring. So scalable video coding (SVC) is presented. Recently, it has been researched intensively and has got noteworthy achievement.This thesis makes a fairly in-depth study on the basic techniques about the current SVC algorithms by reading a lot of relevant references, including the SVC system schemes and the techniques of various scalabilities.Based on these knowledge, Motion compensated temporal filtering (MCTF) which is an effective method to eliminate temporal redundancy is researched , and MCTF based on db2 wavelet is presented by studying the method of MCTF using Haar wavelet. The simulation proves that the low-pass frame is better than the Haar one, it means that the temporal scalability is improved.After completing the MCTF using db2 wavelet, it is well integrated with discrete wavelet transform(DWT) and embedded zero tree wavelet. It uses Atom structure to organize the coded bit-stream to achieve the brilliant combination of three scalabilities: temporal, spatial and PSNR scalabilities. And the software platform is based on VC++6.0.The experimental results manifest that the PSNR of both the full reconstructed video sequences and the ones of frame rates changed are improved compared to the system base on Haar wavelet. Moreover, the decoding time of db2 filters is shorter than the Haar one, the shorter the decoding time, the better the real-time displaying in the client side.