A Reconstruction Algorithm with Adaptive Weighted Fusion in Compressed Sensing

An efficient energy-saving method is proposed by taking the advantages of lightweight encoder and high compression efficiency of Compressed Sensing, which is extremely suitable for image nodes in Wireless Multi-media Sensor Networks. However, the Orthogonal Matching Pursuit (OMP) reconstruction algorithms in CS can not balance the reconstruction image quality and processing time. A novel reconstruction algorithm is proposed in this paper. This algorithm fuses images which reconstructed by OMP according to the row vectors column vectors and diagonal vectors of sample data. And the fusion weights are adjusted by calculating the PSNR values among RIRV (Reconstructed Image by Row Vectors), RICV (Reconstructed Image by Column Vectors) and RIDV (Reconstructed Image by Diagonal Vectors). Experiments show this algorithm is workable, and it reduces the stripe noise. Also, it has a better performance of balancing the reconstruction image quality and processing time than traditional OMP and CoSamp algorithms.