Lossless Image Compression Exploiting Photographic Image Characteristics

Abbasi, I.

Abstract

Digital Images require large storage space as higher and higher resolution becomes possible, at the same time as the storage becomes cheaper it becomes feasible to store rather than discard useful detail. Compressing digital images not only saves storage space but reduces transmission time if the image has to be transmitted. Depending on requirement images are either compressed using lossy or lossless compression methods. Lossy methods allow very large compression ratios as compared with lossless compression methods at the expense of losing information. In cases where smallest image detail matters such as in medical image processing, preservation of art work and historical documents, satellite images and images from deep space probes images are compressed using lossless image compression methods. Despite the importance of lossless image compression of continuous-tone images there is a paucity of standard algorithms. This thesis analyses different methods of lossless image compression which use prediction based on context. These methods exploit information from context and the performance of these methods is proportional to the precision of prediction. There may be more room for compression because more information can be exploited from the context. It is demonstrated that combining existing methods or including more information from the context can improve prediction results thereby improving compression ratios. Results are compared with JPEG-LS which is state of the art method in Lossless Image Compression.

The thesis is available as PDF (6.8MB).