基于余弦拟合和变指数的p-Laplace正则化的活动轮廓模型

Active Contour Model Based on Cosine Fitting and Variable Exponent p-Laplace Regularization

利用余弦拟合能量作为传统活动轮廓模型的数据拟合能量项,并结合变指数的p-Laplace方程,提出了一种新的基于区域的活动轮廓图像分割模型.该模型可以分割复杂图像,变指数的p-Laplace函数可以有效的处理有复杂拓扑变换的图像和准确的提取目标边界.试验结果表明,该模型能够快速准确地分割复杂图像,并且对噪声有一定的鲁棒性和对轮廓的初始位置不敏感.

In this paper,we propose a new region-based active contour image segmentation model,which use the cosine fitting energy as the data fitting energy term for the traditional active contour model,and combined with the variable exponential p-Laplace equation.The proposed model can segment complex images,and the variable exponent p-Laplace function can effectively process images with complex topological transformations and accurately extract target boundaries.The results of experiments show that the propsed model can segment complex images quickly and accurately,and is robust to noise and insensitive to the initial position of the contour.