关键词: 图像分割, 水平集, 灰度不均, 主动轮廓, 偏移场校正, 变分法

Abstract: Intensity inhomogeneity seriously affects the accuracy of image segmentation and active contour model is widely used in image segmentation. In order to overcome the influence of intensity inhomogeneity on image segmentation, this paper puts forward a kind of active contour model based on variational level set. The proposed model uses the mean and variance of the local statistical information of the image, which is suitable for the segmentation of intensity inhomogeneity image. To test the performance of the algorithm, the proposed level set method and the classical algorithm are compared. The results show that the method is validity with both synthetic and real image, and in the process of curve evolution re-initialization is unnecessary, so the method reduces the amount of computation to a certain extent.

Key words: image segmentation, level set method, intensity inhomogeneity, active contour, bias field correction, variational method