Multiprocessor Image Rotation

A technique for reducing the time taken to perform image rotation on general purpose computing hardware is presented. A powerful attribute of the technique is that it is easily mapped onto multiprocessor systems, returning a performance proportional to the number of processors employed. In this article, particular attention has been given to the issues of data management and data communication associated wtih a multiple-processor approach to pure rotation of images. The speed enhancement obtained using concurrently operating processors has been quantified and the limitations have been examined. With the performance achieved, low-cost PC-based multiprocessor systems can satisfactorily process the satellite data now readily available from National Remote Sensing Centres.