Research On Image Segmentation Method Based On Active Contour Model And Level Set

Image segmentation is one of the research hotspots in the field of computer vision,which can be applied to target recognition,medical imaging and image analysis.So far,there is no method to effectively segment all kinds of images,and the existing image segmentation methods still have some problems such as poor segmentation accuracy,low efficiency and poor universality.Therefore,the study of image segmentation is of great significance.Image segmentation based on ACM and level set method has attracted the attention of the researchers due to its flexible topological structure.In the paper,the image segmentation algorithms of active contour model and level set method are studied in depth.The main research contents are expressed as follows:Firstly,the building principles of parametric and geometric active contour models are introduced,the curve evolution and numerical realization of the level set theory are elaborated in detail,and the two evaluation indicators for the precisions of image segmentation algorithms are introduced to provide an objective reason for evaluating the performance of subsequent algorithms.Then,the two edge-based active contour models are summarized,the paper introduced the principle of the models,implementation processing,and carries on the simulation experiment of the kind of model.The advantages and disadvantages of these models for different types of image segmentation are analyzed combined with the experiment results.Secondly,aiming at the problem that the global active contour model cannot segment the images with intensity inhomogeneity and the local active contour model is sensitive to the initial contour,a novel CV model based on improved local information for image segmentation is proposed.The generalized gaussian kernel function is added to local information in the method,the controllable decay rate is introduced to smooth noise,combining global and local intensity information to form a new intensity fitting function,the final level set function is derived from evolution equation of the level set,and the zero level set function is extracted as the segmentation result of the object.The proposed method comparisons with CV,LBF and LCV models in segmentation results of images with intensity inhomogeneity,synthetic images,multi-object and infrared images,and calculation results of two objective evaluation criteria,it can be seen that the proposed method has the advantages of high precision,wide range and strong robustness to initial contour.Finally,considering the edge-based active contour model is easily affected by the gradient information and local binary fitting model is easy to produce the phenomenon of over-segmentation and slow processing,the paper proposed a fast image segmentation algorithm based on edge and local intensity fitting information.The algorithm combines the advantages of the edge-based model and local active contour model,based on the idea of the secondary segmentation,using DRLSE model to rapidly arrived to the object contour,the segmentation result of DRLSE as the initial contour of local intensity fitting model,and then get the more accurate segmentation results.In this way,the contour can be accurately evolved and the process to rough segment using the edge-based model is faster,the segmentation results using local intensity fitting model are more accurate.By comparing the DRLSE model and LBF model segmentation results and calculating the iteration numbers and computation time of three models,it can be observed that the proposed model requires less processing time than DRLSE and LBF models to ensure segmentation accuracy,which is an efficient and fast image segmentation algorithm.