基于局部熵拟合能量与全局信息的活动轮廓模型

Active Contour Model Based on Local Entropy Fitting Energy and Global Information

为解决活动轮廓模型在分割灰度不均匀的图像时存在的对初始轮廓位置敏感、参数选取繁琐及迭代次数多等问题,构建一种基于局部熵拟合能量与全局信息的改进活动轮廓模型。选取灰度图像的中心作为水平集初始轮廓的中心点,改变轮廓半径的大小以确定初始轮廓的位置。使用局部熵项来增强图像边缘处的响应,将局部熵图像拟合能量项与RSF模型共同构成局部能量项,并引入图像的全局信息来避免陷入局部极小值。在此基础上,修正正则项中的长度项,以提高图像分割的效率。在灰度不均匀的合成图像及真实医疗图像上的实验结果表明,与CV模型、RSF模型相比,该模型在进行图像分割时迭代次数较少,精度较高。

Active contour model is very important in image segmentation.However,when dealing with images with intensity inhomogeneity,this model is sensitive to the initial contour position,and its cumbersome selection and multiple iterations can also cause problems.To address these problems,this paper proposes an active contour model based on local entropy fitting energy and global information.Firstly,we select the intensity image center as the central point of the level set of initial contour,and change the radius size to determine its location.Then,we use the local entropy to enhance the response from the image edge,and combine the local entropy image fitting energy and RSF model together as local energy terms.Besides,the global information of image is added to avoid falling into local minimum.On this basis,we amend the length in the regular terms to improve the segmentation efficiency.Experiments on synthetic images and real medical images with intensity inhomogeneity show that the proposed model has fewer segmentation iterations,and the results are more accurate.