Parallel JPEG algorithms for still image compression

Abstract

One of the hottest areas of development in the computer industry today is motion video on the desktop. Both motion video and still image video pose some new problems: huge storage and high bandwidths are required to send, view or store video data in a timely fashion. Compression techniques may help to reduce the amount of data by trading processing power for reduced data size. For both video compression and decompression, very large amounts of processing power are required to handle the video data quickly enough. Parallel processing is being used in some cases to solve this need for processing power. This paper explores some of the parallel processing techniques currently being used, and some additional possibilities for parallelization. It focuses primarily on JPEG (Joint Photographic Experts Group) techniques, but many of the concepts are applicable to other compression schemes. Results from a simple PC-based JPEG compression/decompression program are used to explore some of the techniques.<>