Color Image Restoration And Fusion Via Compressive Sensing

Image fusion integrates the complementary information from various sensors toobtain more accurate, efficient and comprehensive description of the scene of interest.Currently, image fusion has been widely used in medicine, remote sensing, industrial,computer vision and so on. In addition, limited by the hardware conditions, the sourceimage acquired usually disturbed by noise. Therefore, it is necessary to restore thesource image first to improve image quality. The traditional algorithm deal with imagerestoration and image fusion separately. Such algorithms are of low calculationefficiency and fusion results need to be further improved. So, It,s theoretically andpractically important to do research on algorithms deal with image restoration andimage fusion simultaneously.This paper researches the compressed sensing and sparse theory, and apply to colorimage restoration problem. Furthermore we combined color image fusion andrestoration by studying the fusion on the sparse domain. The main work is as follows:We propose a Extended Joint Sparse Model to efficiently exploit the correlationamong different color channels of the color image and improve the performs of thealgorithm proposed in the paper. The model is based on the Joint Sparse Model. Weachieved the color image recovery with the compressed sensing theory by researchingthe characteristics of the Gaussian noise, impulse noise and the demosaic problem. Onthe study of the intrinsic link of the image fusion and restoration on the compressedsensing domain we combine the color image fusion and restoration taking the sparserepresentation as the bridge. Experimental results demonstrate the effectiveness andfeasibility of the proposed method.