Abstract: In the field of image segmentation, geometric active contour model is one of the most successful methods. Most of the existing geometric active contour models need to define an initial position for the evolution curve. This may cause the segmentation results to be affected by the position of initial contour. To this end, a new signed pressure force function is constructed based on the local and global information of the image, and a fast image segmentation model in the form of partial differential equation is proposed. The proposed model is simple and the algorithm process is easy to implement. Experimental results show that the proposed model allows constant initialization. Without initial contour, it can rapidly segment three-phase images, images with intensity inhomogeneity, images with gradually varied intensity, deep images, and so on.

Key words: image segmentation, active contour model, partial differential equation, constant initialization, signed pressure force function

摘要： 在图像分割领域中，几何活动轮廓模型是较成功的方法之一。但现有的几何活动轮廓模型大都需要为演化曲线定义一个初始位置，这容易导致图像分割结果受初始轮廓位置的影响。为此，结合图像的局部和全局信息构造一个新的符号压力函数，提出一个以偏微分方程形式存在的快速图像分割模型。所提模型形式简单，算法过程容易实现。实验结果表明，该模型允许常值初始化，无需初始轮廓即可快速分割三相图像、灰度不均图像、渐变图像以及深度图像等多类图像。

关键词: 图像分割, 活动轮廓模型, 偏微分方程, 常值初始化, 符号压力函数