Fast Evolutionary Image Processing Using Multi-GPUs
Jun Ando, Tomoharu Nagao · InTech eBooks · 2009
We employed GPU to ACTIT for the purpose of reducing optimization time. The proposed method essentially performed the process that costs large computing time on GPU. Moreover, we proposed an efficient parallel model and instructed GPU-ACTIT to perform in parallel using multiple GPUs for fast processing. The optimization of the proposed method was several hundred times faster than that of the ordinary ACTIT. We experimentally showed that the proposed method is effective. In future work, we plan to implement complex filters for ACTIT using only CPU that can be implemented on GPU. We propose image-processing algorithms that are effective for GPGPU. We also aim to construct a fast evolutionary image-processing system.