Image Segmentation Method Based On Geometric Active Contour Model Is Studied

Image segmentation is a key problem of the image processing and the computer vision.The geometric active contour model for image segmentation is a classical method of which the active contour model without edges(the C-V model) has been a classical one. However, when the classical active contour model without edges is used for image segmentation, only the homogeneity information is taken into account, which causes the active contour model without edges to work only images with two inhomogeneous areas but often fails to correctly segment images with inhomogeneous intensity.The prime task of this thesis is as follows:(1) A new similarity based on the Chebyshev distance was introduced into the active contour model without edges and the mean value definition of the whole image in the traditional active contour model without edges model have been changed.The new model solves the problem that the traditional active contour model without edges model can not correctly segment images with inhomogeneous intensity. Experiments and contrast analysis evidently show:the new model can be used to segment images with inhomogeneous intensity, and is not sensitive to the noise.(2)A two-stage method for image segmentation by combining the Top-Hat operator and the geometric active contour model was proposed.The method firstly do the Top-Hat operation for image to strengthen its object meanwhile restrain the background,secondly the improved active contours model without edges based on Chebyshev distance was used to segment the image which has been processed by the Top-Hat operator. Experiments evidently show:this new two-stage method works fastly and accurately to segment some natural images and medical images whith complex background and uneven optical areas.