To address the low-efficiency issue of the traditional multi-resolution image fusion algorithms executing on CPUs
a novel image fusion rule for gradient-pyramid-decomposition based image fusion algorithms is proposed along with the design and efficient implementation of the entire fusion algorithm on Graphics Processing Units (GPUs).The computation-intensive tasks are offloaded from the CPU to the GPU.In the experiments
the fused image preserves the salient features of the source images.Moreover
the GPU accelerated fusion system gains speedups on images of different dimensions versus the pure CPU implementation.Thus an approach to improve the efficiency of multi-resolution image fusion algorithms is presented.