Image Compression Via Sparse Representation

Abstract

In recent years sparse representation has become a hot topic in the research of image representation model and there has been a growing interest in the study of image compression based on sparse representation. However, when coding different images, we have to train the dictionaries that correspond to these images. In the paper, we use three different kinds of image databases to train a group of bases which respectively reflects the characteristics of the three kinds of image, these bases are formed a big over-complete dictionary. Experimental results show that the proposed dictionary is superior to image compression based on the dictionaries trained by single image database.