A low-complexity stereo video coding scheme for surveillance

Low-complexity stereo video compression is important for surveillance. The existing H.264-based stereo video coding compression schemes have high coding performance, but the complexity of motion and disparity estimation is a significant burden for surveillance, especially in the 3D case. In this paper, a low-complexity stereo video coding scheme is proposed for 3D surveillance. This scheme is based on motion vector (MV) extrapolation and region-of-interest (ROI) image restoration technique. In one hand, MV extrapolation refinement is used to reduce the complexity on motion estimation. In addition, through the object segmentation in low-resolution video and information transmission in ROI enhancement layer, the quality scalability for reconstructed frame is implemented. The experimental results show that our proposed scheme can reduce the coding complexity of stereo video while maintaining a high compression performance.