Parallel Wavelet-Based Image Segmentation Using MPI

In recent years,how to do segmentation to large texture image effectively and in high speed has been emerging as an important reseach focus in the field of image segmentation. Finding more useful texture segmentation methods and the research about the parallel image processing are regarded as a solution to the problem.This paper takes the multi-resolution texture image segmentation modeling and its feature attrachtion methods as research objects.trying a texture image segemantaion methods based with wavelet approach and FCM cluster methods.First,this paper make in-depth reasearch on using wavelet transfer to effectily compute the energy of the texture image,secondly doing the research of application of the FCM cluster algorithms,at last design a segmentation framework based them.Considering that wavelet-based image segmentation has very high computational complexity, especially for large images,in order to speed up the process of the large images segmentation, this paper put emphasis on exploring the computational parallelism in wavelet-based image segmentation.The parallelism of the MPI are in-depth analyzed,data parition of the large image also are designed for reducing the data transmission,and parallel wavelet transform algorithms are designed.This paper implement a parallel wavelet-based image segmentation using MPI on cluster machines with the better segmentation rate.