Stereo Video Coding based on Depth Map and SPIHT

In this paper, we propose a scalable stereo video coding algorithm based on Depth estimation and Set Partitioning in Hierarchical Tree (SPIHT) codec. The new approach which combines depth estimation algorithm based on markov random field (MRF) and the wavelet transform technique makes it possible to realize scalable wavelet stereo video coding. Compared to traditional stereo video coding, we use depth estimation to get the depth map and detect the depth of different objects, the wavelet transform coefficients of near object in SPIHT codec will be classified in the significant sets and others insignificant. After SPIHT coding, the wavelet transform coefficients are divided into multiple independent sub-streams for transferring. The nearest of object is often the most frequent visitors of regional concern, so more details are reserved. Finally, we make some assessments to this algorithm in reduction ratio and visual effect. Simulation results on different video sequences show that the proposed method realizes low encoding complexity at the encoder and achieve efficient performance of stereo video compression.