Adaptive lossless image compression is one of the most important applications in the field of evolvable hardware (EHW). However, related studies in the past focused on implementations with extrinsic EHW, which uses a host computer to run software simulation and compiling, and then download the final circuit to the silicon chip. This is not suitable for tasks of on-chip adaptation. This paper presents a novel technique to reformulate the problem as a task of evolving a set of switches. As a result, the whole scheme can be implemented easily using intrinsic EHW. In order to enhance the scalability of the whole scheme, a strategy based on data-decomposition and pyramidal fitness evaluation strategy is developed for evolving larger scale images. Software simulation shows that the proposed method can largely reduce the computation time, and can scale up the image size up to 70 times with relatively slow increase in computation time. Hardware simulation shows that the method can be applied in practice.
Bibliographical noteThis work is partially supported by the National Natural Science Foundation of China through Grant No. 60573170 and Grant No. 60428202.
- Evolutionary computation
- Evolvable hardware
- Lossless image compression