Cost-sensitive Information Maximization and Its Application to Image Compression

R. Kamimura and O. Uchida (Japan)


mutual information maximization,competitive learning, cost, image compression


In this paper, we apply cost-sensitive information maxi mization to image processing. Cost-sensitive information maximization aims to increase information content while the associated cost is controlled to be minimized. This method is called cost-sensitive information maximization. We tried to apply the method to image compression. In im age compression, one of the most important things to do is to restore original images as accurately as possible. Thus, cost minimization should be effective in image compres sion. Experimental results confirmed that information can be increased by cost minimization. In addition, cost min imization give better compression performance than con ventional methods.

Important Links:

Go Back