A New Method for Compressed Sensing Color Images Reconstruction Based on Total Variation Model

Abstract

A new method based on the total variation model, applicable to reconstruct the compressed sensing color images, is proposed. At first, the compressed sensing color images should be converted form the RGB color space to the CMYK space, and the compressed sensing color images in the CMYK space can match exactly with the quaternion matrix. Next, the amplitude and the different four phase information of the quaternion matrix is treated as the smoothing constraints for the compressed sensing problem in order to reconstruct the color images more effectively. Finally, the gradient projection method is used to solve the compressed sensing problem. Experimental results show that this new method can reconstruct color images better than some traditional methods.