Some New Kinds Of Active Contour Models And The Realization

With the development of the science and technology and widely usage ofcomputer, technology of digital image processing have come into everyday lifeand have an important effect on everybody. In recent years digital imageprocessing becomes one of hot-topics in the field of scientific researchincreasingly.The Active Contour Model is the key representation of the method.Based on literature in existence, this paper induces Image gravitation, PSO,PCNN, HMM, and does some works as follows:(1) The concept of the image gravitation is proposed. The experimentshows that there is a good feather in the model.(2) A new active contour model based on Fuzzy and HMM is provided.HMM is used in the active contour model and the parameters of the model isoptimized using the ability of the measuring many kinds of information basedon the HMM. In additional, the Fuzzy theory is used to make the relationshipbetween Fuzzy membership function of the state and observed vectors in theHMM. The experiment shows that the boundary of the hollow objects can bereached better in the active contour model which is improved, and there is agood anti-noise feather in the model.(3) A new active contour model based on image gravitation and PSO isprovided. The concept of image gravitation is used in the model to make thesnake point reach the edge of the image better. In the other hand, the method ofPSO is also used which making the snake point reaching the edge of the objectexactly to prevent the snake point being affected from the noise. Theexperiment of the dynamic image shows that the boundary of the objects can bereached faster and better in the active contour model which is improved,especially the hollow objects, and there is a good anti-noise feather in themodel.(4) A new active contour model based on PCNN is provided, which iscalled the PCNN-Snake. PCNN is used in this model to avoid the snake pointreach the noise point. PSO is also used in this new model to select the thresholdof the PCNN-Snake, which can make the snake point reach the edge better. Theconcept of the image barycenter can make the snake point reach the edge of thehollow edge. And the integration of the image barycenter and the PCNN canmake model not only reach the hollow edge but also avoid reach the noise point.The experiment shows that the boundary of the hollow objects can be reachedbetter in the new model , and there is a good anti-noise feather.(5) Releasing it by VC.