The Design And Optimization Of Scalable Video Encoder Based On Wavelet Transform

The Wavelet Transform, a new kind method of transform, developed in recent years. The study on Wavelet Transform has converted many fields such as mathematics, physics and engineering and has become a popular field of information technology. The case of Wavelet Transform so popular is that it has good localized capability in time and frequency field simultaneously. And since we can take gradual Q_f in the high frequency subband of time or frequency field, so the detail of the signal can be detected. The Wavelet Transformed coefficients has two main characters: combination with HVS and multi-resolution. Under the status of the violent development of multimedia and the heterogeneity of network, application and user demand, the research on the scalable video coding based on Wavelet Transform is necessary.The paper introduce the principle of Wavelet Transform and Scalable Video Coding firstly, then the Dirac Video Codec based on Wavelet Transform which is proposed by BBC RD and its comparison with X264 and SNOW. The primary achievements of the dissertation include: A temporal and spatio scalable coding scheme based on the hybrid 2D Wavelet Transform and Motion-Compensate structure, and the optimization of it. Using the IA MMX instructions to help the bottleneck functions with large numbers of repeating calculations executing parallelly. The result of Intel Vtune Performance Analyzer 8.0 proves the performance of these functions has been improved well.