关键词: 图像分割, 水平集, 曲线演化, 偏微分方程, 活动轮廓模型

Abstract: It is difficult to get the correct segmentation results for the intensity inhomogeneity images, and the segmentation results are very sensitive to the initial contours. Thus, a fast and stable approach was proposed to overcome these disadvantages. First, an Adaptive Distance Preserving Level Set (ADPLS) method was utilized to get a better initial contour. Second, the Local Binary Fitting (LBF) model was used for a further segmentation. The experimental results show that the improved model can achieve good performance and is better to solve the contradiction among the segmentation speed, accuracy and stability.

Key words: image segmentation, level set, curve evolution, Partial Differential Equation (PDE), active contour model