Optimization Techniques to Reduce Computational Complexity for Scalable Video Coding

Scalable video coding (SVC) is an ongoing video coding standard that is an extension of H.264/advanced video coding. The goal of SVC is to provide scalability at the bitstream level with good compression efficiency by allowing combinations of scalable layers, including the spatial layer, temporal layer, and signal-to-noise ratio layer. This enables a simple and flexible solution for transmission over heterogeneous networks. Coding efficiency in SVC has significantly improved; however, an extremely high computational complexity is still required in determining the optimal coding mode for each macro-block. Coding modes are selected by using variable block-size motion estimation. Therefore, it is desirable to develop optimization algorithms to reduce the computational complexity of SVC without compromising the coding efficiency. In this project, three research directions are proposed to reduce the computational complexity of SVC while keeping the coding efficiency unchanged:explore the mode-distribution correlation in SVC to build mode prediction models for shrinking the mode selection range;present advanced techniques to develop fast-mode decision and motion estimation algorithms to reduce the computational complexity of SVC; andconsider the optimizing of the discrete cosine transform and quantization functions to further reduce the SVC computations.