| Image segmentation and boundary extraction are very important in the fields of image understanding, pattern recognition, computer vision and so on. They are also the basis of the image analysis. Up to now, there is still not a common method for any type of images in image segmentation. Recently, image segmentations based on partial differential equations (PDE), one of the novel and efficient segmentation methods, are gradually turned into research hotspot. Geometric active contours, i.e., active contour implemented via level set methods, have been proposed to address a wide range of image segmentation problems in image processing and computer vision.After reviewing the literatures involved region-based segmentation methods, this dissertation discusses the Chan-Vese model and RSF (Region-Scalable Fitting) model and obtains the following results:1) The Chan-Vese model based on global region information is less sensitive to initialization and noise; however, it cannot handle images with intensity inhomogeneity. The RSF model based on local region information is able to deal with intensity inhomogeneity; however, it is highly sensitive to initialization and noise. In order to address this problem, this dissertation proposes a new region-based active contour model, which integrates both global and local region information. Experiments show that the proposed model can segment images with intensity inhomogeneity, while it allows for flexible initialization and is less sensitive to noise.2) The RSF model is able to deal with intensity inhomogeneity; however, it is sensitive to initialization and noise. In order to address this problem, this dissertation proposes an improved version of the RSF model by adding an energy functional of level set linear regularization into RSF model. The proposed model can address the segmentation of images with intensity inhomogeneity, while it allows for flexible initialization of the contours and is significantly less sensitive to noise. |
PDF Full Text Request