Automated Design of an ASIP for Image Processing Applications (Research Note)

This paper presents the design of highly optimized TTA architectures for image processing applications. An automatic processor design framework as described in [2] is used. Specialized hardware is used to improve the performance-cost ratio of the processors. An explorer searches the design space for solutions that are good in terms of cost and performance. We show that architectures can be found that efficiently execute very different algorithms at low cost. A hardware feasible architecture is presented that efficiently executes a set of image processing algorithms and performs almost equally or better than alternative, commercial-available solutions do.