Hybrid Active Contour Model Integrating Intensity And Differential Information

Image segmentation is an important research in the field of computer vision and image processing. Until now, there is a variety of segmentation methods, and among them active contour model is a popular one widely studied and used in many applications. The fundamental idea of active contour model is to construct some constraints to the object and realize the segmentation process by curve evolution.This paper is mainly research on the method of the active contour model. Our paper introduces several kinds of active contour models, and makes a brief analysis of the advantages and disadvantages of the model according to the experimental results. The experiments show active contour model which use the image global information is not sensitive to the initial contour, but it can’t accurately segment the images with intensity inhomogeneous, while the active contour model which use the image local information can effectively deal with the images with intensity inhomogeneous, but it is sensitive to the initial contour. In order to achieve a more effective segmentation of the image, a hybrid active contour model integrating the differential and local gray intensity information is proposed in this paper, where the differential information of image is that the original image from a averaging convolution operator subtract original image, and the time step method is adopted in the numerical implementation, and the process of re-initialization in the traditional update iteration is avoided, so the numerical implementation is more simple in this paper. In addition, based on the model, a hybrid active contour model integrating background estimation information is proposed in this paper. Where the background estimation information is that the background image subtract original image, and the background image is the result of morphological open operation on the original image. The data term in the energy functional of proposed model consists of background estimation information and local information of image. Where fitting the image local information can maintain the model accuracy of image segmentation, fitting the background estimation information can overcome the problem of initial contour effectively. And background estimation information itself can description of the images with intensity inhomogeneous, so our model avoids the problem which exists in the hybrid model that the parameter of local term and global term is difficult to select an appropriate value. Experimental results show the proposed model can obtain improved segmentation result compared to related methods in both segmentation accuracy and initial contour sensitivity.