A Scalable Block Compensation with Index Selection Mechanism to Achieve High Video Quality Retrieval

Video quality retrieval is an important issue in reconstruction video sequences. The amounts of transmission bitstreams are very large to achieve high video quality reconstruction. Thus, high video quality retrieval is still a challenge problem. The aim of this paper is to reduce the transmission bitstreams and raise reconstruction video quality. In this paper, we propose a scalable block compensation with index selection mechanism using eigenvalue and local variance statistics for each group of pictures (GOP), which are regarded as the feature parameters for compensating motion blocks for each video frame from low-resolution to high-resolution sequences at the receiver ends. To reduce transmission overhead, the index selection technique is then applied to those selective compensation blocks to have variances larger than a threshold. The results of index numbers from index selection are regarded as reconstruction information. Simulation results show the percentage of noncompensation blocks is nearly 30% saving for Mobile, 36% saving for Foreman, and 48% saving for News. These results indicate that the proposed algorithm offers a more efficient way to decrease the percentage of compensation blocks about 30~48%. The proposed algorithm, on average, outperforms H.264/AVC about 2.8~9.8 dB from 100 kbps to 1 Mbps.