C-V模型的一种集成化改进

AN INTEGRATED IMPROVEMENT OF C-V MODEL

C-V(Chan-Vese)模型不能分割灰度不均图像,分割速度慢。因此,对图像分割理论进行深入探讨,采用实验方法,提出一个新的边缘停止函数,并将其与图像灰度变换及H1正则项融入C-V模型中,对其进行集成化改进,提出一个新颖的以偏微分方程形式表示的活动轮廓模型。实验表明:该模型能分割灰度不均图像,对初始轮廓的大小敏感性较小,抗噪性较强,分割速度较快。

C-V （Chan-Vese） model can not segment images with intensity inhomogeneity, the segmentation speed is also slow. Therefore we discuss in depth the image segmentation theory and propose a novel edge stopping function with the experimental method, and integrate it to C-V model together, with images gray-scale transformation and Hl regular term. The algorithm is also made the integrated improvement, and a novel active contour model expressed in partial differential equation （PDE） is proposed. Experimental results show that the improvement model can segment images with intensity, inhomogeneity, it is less sensitive to the size of initial contours and is significantly less sensitive to noise, and its segmentation speed is faster as well.