Study On Image Segmentation Based On AP Clustering Algorithm

Clustering analysis is an important technology in data mining, and it has been widely used in areas such as statistics, image processing,medical diagnosis,information retrieval,biology and machine learning. The clustering algorithm can achieve a better effect of segmentation in the process of image segmentation, caused the wide attention of experts and scholars. Image segmentation is an important step of image processing analysis and image recognition, but also makes use of computer to realize the core link of low-level to high-level visual.With the development of artificial intelligence, image segmentation techniques are extensively used in varieties of subjects such as image guidance, unmanned aircraft, vehicle license plate recognition, medical image recognition, voice recognition, face recognition, fingerprint recognition, currency recognition. The cores for implementing them are picture processing, computer vision, pattern recognition, intelligent control and other related knowledge. It is visible that the quality of picture segmentation directly influences the subsequent image processing. Moreover, it even determines its success or failure. Therefore, segmentation methods and the accuracy are the key procedures from image processing to image analysis. It is as well as a significant issue for picture processing, pattern recognition and artificial intelligence.This thesis, based on the overseas and domestic research status, setting about the theory of picture segmentation, elaborates from varieties partition methods such as traditional picture segmentation methods(threshold segmentation algorithm, edge segmentation algorithm, and region segmentation algorithm), and the methods based on special theories(partition methods based on fuzzy theory, level set, active contour models and graph theory). And then the author solves the problem through clustering algorithm. He specific describes the principles, implementation methods, merits and demerits of K-means arithmetic, hierarchical clustering arithmetic, spectral clustering arithmetic, as well as mean shift arithmetic. He also put forward the conception of “Picture segmentation technologies based on AP clustering algorithm”, and he also makes simulation for partitioning algorithm through MATLAB, which finally solve the problem of picture segmentation. This thesis, the author hopes, could promote the maturation and development of picture segmentation, providing a significant enlightenment for solving other similar complex pattern recognition issues.