A parallel image fusion algorithm based on wavelet packet

The wavelet packet provides an accurate method for image fusion. However, the computation and time cost will increase with the size of image. Therefore it is difficult to achieve real-time fusion. After having analysed the wavelet packet based image fusion algorithm on the single computer, according to the time complexity and the parallel character of the wavelet packet algorithm, we propose a wavelet packet based parallel image fusion algorithm under the situation of distributed storage. Based on the data local property of wavelet packet transform, we achieve the algorithm on the MPI (Message Passing Interface) parallel computing platform, which is composed by Pentium PC and 1000 Mbps Ethernet. The efficiency of parallel computing is studied under different image size and different cluster size. The experimental result shows that the algorithm has good parallel computing property.