Research On The Application Of Active Contour Model In Image Segmentation

Among the theories of image segmentation,the active contour image segmentation method based on variational level sets is one of the hot research topics.In this paper,we start the research from two active contour models respectively,and propose new models for segmenting noisy images and the problems encountered in image segmentation by mixed global and local active contour models.The research work is as follows:1.An active contour model based on locally fitted division is proposed.The energy generalization function is established by building a local division function,extracting image information within the local object and local background respectively,and finally using finite difference and gradient descent methods to achieve numerical computation.The experimental results show that the proposed model has good robustness to noise in images and achieves good Dice Similarity Coefficient(DSC)values,which indicates that the model has some robustness to noise.2.A global division active contour model based on fusing local information(DGLF)is proposed.The energy generalization function in the(An Intensity-Texture model based level set method for image segmentation,ITLSM)model is linearly combined with the energy generalization function in the Local Binary Fitting energy(LBF)model,and in addition,the energy generalization function in the(An Intensity-Texture model based level set method for image segmentation,ITLSM)model is linearly combined with the energy generalization function in the Local Binary Fitting energy(LBF)model.An adaptive function is established as the weight coefficients based on the experimental maximum number of iterations to achieve the adaptivity of the weight coefficients.The experimental results show that the DGLF model can segment some more complex images and achieve better Jaccard Similarity Coefficient(JSC)values,which further illustrate the segmentation performance of the DGLF model.