结合全局和局部信息的水平集图像分割方法*

LBF模型对初始轮廓大小和位置非常敏感，并且只考虑了图像的局部信息，没有考虑图像的全局信息。CV模型利用图像全局信息，对初始轮廓具有较强的鲁棒性。两种模型对椒盐噪声污染的图像不能取得令人满意的结果。针对以上问题， 在原有CV模型和LBF模型能量函数基础上，各自构造一个新的能量拟合项，增强对高斯噪声和椒盐噪声的抗噪性。采用新构造的CV模型，使用图像的全局信息得到粗分割轮廓。以粗分割轮廓作为新构造LBF模型的零水平集，利用图像的局部信息得到图像的精确分割结果。同时提出一种新的边缘检测算子，重新定义边缘停止函数，进一步提高模型的抗噪性。相较于CV模型，LBF模型，结合全局和局部信息的Wang模型和Qi模型，提出模型能得到更优的图像分割结果，具有较强的抗噪性。

Level set image segmentation method by global and local information

LBF model is very sensitive to the size and location outline, and it only considers the local information ,without considering the global information of the image. CV models considers global image information, it is robust to the initial profile. LBF and CV model cannot obtain satisfactory segmentation results for Salt and Pepper noise pollution image. to solve these problems, a new energy fitting items is defined respectively based on the original CV model and LBF model energy function to enhance noise immunity for Gaussian noise and Salt & Pepper noise. the improved CV model is employed to obtain coarse segmentation, the improved LBF model is employed to obtain accurately segmentation result with the initial contour based on the coarse result. A new edge detection operator is proposed to redefine the edge stop function to further improve the noise immunity of the model. compared to the CV model, LBF model, Wang model and Qi model using global and local information, the proposed model can get better segmentation results, with strong noise immunity.