Methods for Lossy Compression of Images Corrupted by Multiplicative Noise

Compression of images corrupted by multiplicative noise is characterized by some specific features. First, lossy image compression is preferable since lossless coding is ineffective. Second, in case of lossy compression, quality of a compressed noisy image evaluated with respect to noise-free one can become better than the corresponding quality of an original image. And there exists such compression ratio called optimal operation point for which the quality of lossy compressed image is the best. Below typical situations that can be observed in compression of remote sensing data, for example, radar images are considered. We evaluate the efficiency of two basic methods for lossy compression of images corrupted by pure multiplicative noise. A way to reach optimal operation point for providing the best quality of coded images is proposed. The corresponding automatic procedure is described.