Intelligent Image Segmentation Methods Based On Variable Precision Rough Entropy

Image segmentation is not only a basic issue but also a key step from image processing to image analysis in image engineering. It makes possible further higher level image processing such as feature extraction, classification, pattern recognition and retrieval in computer vision. Therefore, high quality and robustness of image segmentation routines is of considerable impor-tance in the practical application. Thresholding is one of the old, simple and popular techniques for image segmentation and has been intensively studied.However, it is difficult to select a proper threshold for image segmentation because that various regions have fuzzy boundaries, nearby gray levels roughly resemble each other and values at nearby pixels have rough resemblance. With the development of intelligent computing, lots of new theories and new methods of image segmentation come out. In recent years, the rough set theory and fuzzy set theory have been introduced to the image processing. Utilizing rough entropy or fuzzy entropy has made much progress in image segmentation.On the basis of the theory of image segmentation, thresholding images based on rough entropy is studied in this thesis deeply. Combining with variable precision, fuzzy logic and intelligent optimization algorithms, several image segmentation algorithms are proposed for different thresholds requirements. The main contribution in this thesis can be summarized as follows.(1) Introduced the notion of inclusion measure, a representation model of image by vari-able precision rough set is proposed. In combination genetic algorithm with rough entropy, a new image single threshold value segmentation algorithm based on variable precision rough set is put forward.(2) A general representation model of image by rough set band fuzzy set theory is studied. The proposed model first blurred image, then defined fuzzy set inclusion through fuzzy logic operator. According this, image can represent by rough set and rough entropy. The proposed method can not only reduce the impact of noise to the process of image segmentation but also degradation to the classic image representation model of rough model.(3) In combination particle swarm optimization algorithm with roughest representation of image extended to the case of multi-threshold, a multi-threshold image segmentation method is investigated.(4) Experimental results show that the proposed algorithms are more effective and flexible.