Compressive sensing based scalable video coding for space applications

Abstract

In this paper, a compressed sensing based scalable video coding using discrete wavelet transform (DWT) for space applications is presented. The proposed method is aimed at reducing the requirement of on-board memory, station contact time and data archival volume. Compressive sensing (CS) has been applied to encode the distributed wavelet coefficients into linear measurements. Since prior knowledge of the compressively sampled measurements is not available, a modeler is employed to determine the context of the measurement vector. Based on the statistics provided by the context modeler, the samples are entropy coded for transmission. A part of the compressive sampling measurements with near uniform distribution are encoded through adjusted binary coding, and the rest are encoded through Golomb-Rice coding. Run length encoding is implemented for exploiting the redundancy in the binary data further. At the receiver end, the original spatial signal is reconstructed from the CS samples by approximate message passing (AMP) algorithm and inverse discrete wavelet transform (IDWT) for video decompression. Simulation of the proposed approach has demonstrated an efficient compression of the video cubes with lower complexity at the transmitter end.