Statistical Image Modeling And Image Segmentation In Contourlet Domain

Image segmentation is to divide a given image into regions of various characteristics by a certain criteria, extracting the interesting targets. The difficulty of texture image segmentation is that the result of image segmentation is good in region homogenous and edge segmentation at the same time, the methods existed are inaccessibility in these two aspects. According to this, as an important branch of image analysis, image segmentation is always the focus of research in the field of image engineering.In order to process and analyze images effectively, many people give the enormous attentions to statistical image modeling technology all the time. In the image segmentation domain, the statistical image models achieves good application effect; its correlative technique's research becomes the hot spot in the domestic and foreign scholars' study. In this dissertation, statistical image modeling on Contourlet domain is studied in terms of the image segmentation application, and some new methods are proposed. The main contents of the thesis can be summarized as follows:(1) The relevant theory from Wavelet to multi-scale analysis is investigated. The application in image processing of HMT model in Wavelet and Contourlet domain is specially investigated.(2) The algorithm of HMT model in Contourlet domain is specially investigated. A computational method of boundary domain is adapted from the idea of Multi-scale Bayesian fusing strategy combining with the boundaries of Chen Rongwei, which is used to guide the boundary detection, and combining it with Multi-scale Bayesian fusing strategy. The experiments of segmentation in texture image verify the effectiveness of this method.(3) Take full account of advantages in image edge extraction of watershed algorithm, and avoid over-segmentation problem of it, a new SAR image segmentation algorithm is proposed. The algorithm combine the HMT model in Contourlet domain with the watershed algorithm. The experimental results show that the new algorithm has better segmentation in small edge detection and regional consistency.