High-Performance Distributed Compressive Video Sensing: Jointly Exploiting the HEVC Motion Estimation and the ℓ₁ – ℓ₁ Reconstruction

Abstract

The distributed compressive video sensing (DCVS) system combines the advantages of compressed sensing (CS) and distributed video coding (DVC), suitable for the limited-resource video sensing and transmission environment. In this paper, we propose a comprehensive high performance DCVS system. First, we introduce the BM3D-AMP algorithm reconstruct key (K) frames. Second, we propose a new high efficiency video coding (HEVC) motion estimation (ME) algorithm with motion vector (MV) prediction method. By integrating the segmentation idea and motion estimation, this algorithm can gets more accurate side information (SI). Finally, we propose the ℓ1 - ℓ1 minimization model to achieve non-key (NK) frames joint high-quality reconstruction. We utilize the alternating direction method of multipliers (ADMM) algorithm to solve it. With the idea of dividing and conquering, the general problem is decomposed into several smaller pieces. Experimental results demonstrate that the proposed system has significant improvement over its counterparts.

Keywords

• Distributed compressive video sensing
• side information
• HEVC motion estimation
• ℓ₁–ℓ₁ minimization