Skip to Main Content
In H.264/AVC, the discrete cosine transformation (DCT) coefficients of a residual block are often quantized to zeros. Such block is called a all-zero block(AZB). If such AZB can be detected before transformation, the computation of following transformation and quantization can be avoided. This will lead to significant saving both in encoding time. An efficient and simple AZBs detecting algorithm in H.264/AVC based on human visual system(HVS) is proposed in this paper. First, the spatial and temporal features of video pictures are analyzed and four grades based on subjective video pictures quality are set up. Second, a more precise threshold than previous algorithms is obtained by analyzing the theory of integer transform and quantization. As a result, a more efficient and simpler AZBs detecting algorithm based on HVS characteristics is proposed. The simulation results show that the proposed algorithm achieves better performance in encoding time saving and subjective video pictures quality without magnificent PSNR degradation and bitrates increment, compared to the previous algorithms.
Date of Conference: 25-28 Sept. 2011