Abstract: Chan-Vese (C-V) model (CHAN T F, VESE L A. Active contours without edges. IEEE Transactions on Image Processing, 2001, 10(2): 266-277) is one of the well-known region-based image segmentation models. It is much less sensitive to the initialization of the contours and noise; however, the range of segmented images are not enough extensive. Therefore, through using the method of combining theoretical analysis with experiment and adding H1 regular term in C-V model, the algorithm was improved. A novel energy function for image segmentation was proposed, and an active contour model of region-based adaptive interpolation fit for a Partial Differential Equation (PDE) formulation was deduced. The experimental results show that the improved model can segment some images that C-V model is not applicable, and it is less sensitive to flexible initialization contours and significantly less sensitive to noise.

Key words: image segmentation, active-contour, H1 regular term, partial differential equation, Chan-Vese (C-V) model

摘要： C-V模型(CHAN T F, VESE L A. Active contours without edges. IEEE Transactions on Image Processing, 2001, 10(2): 266-277)是一个著名的基于区域的图像分割模型。它对活动轮廓的初始化和噪声不敏感，但分割的图像的范围不够广泛。因此，运用理论分析与实验相结合的方法，在C-V模型中添加H1正则项，对其进行了改进，提出了一个新颖的图像分割的能量泛函，并推导出了以偏微分方程形式表示的基于区域的自适应插值拟合的活动轮廓模型。实验表明：该模型能够分割某些原来C-V模型不适用的图像，它对初始轮廓的大小、位置的敏感性较小，抗噪性较强。

关键词: 图像分割, 活动轮廓, H1正则项, 偏微分方程, C-V模型