Distributed compressed sensing for image signals

Distributed compressed sensing (DCS) is able to exploit both intra-and inter-signal correlation structures of multi-signal ensemble. This paper proposes a DCS scheme for image signal compression and reconstruction. The key idea is to exploit the inter-correlation of the blocks that split from the image. Significantly, joint sparse model was employed to compress the intra- and inter-redundancy of the image signal. Moreover, our scheme allocates more sensing resources to common component while fewer measurements for innovation component. In order to improve the performance, we also utilize variable sizes method to replace the uniform size approach for image split. Experimental results on natural images validate that the proposed DCS scheme validly improves the reconstructed image quality with fewer measurements compared to the existing CS schemes.