关键词: 图像分割, 水平集方法, 活动轮廓模型, 边缘停止函数, 局部熵, 灰度变换

Abstract: Level Set Evolution Without Re-initialization（LSEWR） is a well known variational level set model.It completely eliminates the re-initialization procedure of level set function.However，because its edge stopping function is built based on image gradient，it has still several disadvantages.It is highly sensitive to noise，and is prone to edge leakage when applied to images with weak edges.Finally，it is not available for images with discontinuous edges.A new edge stopping function is constructed based on local entropy and gray-scale transformation.Experiments show that the LSEWR model with the new edge stopping function can do a good work for overcoming the above-mentioned drawbacks.

Key words: image segmentation, level set method, active contour model, edge stopping function, local entropy, gray-scale transformation