纹理图像的相位场演化分割研究*

作者：周佳男1，2，刘均1，冯志林1，赵端阳1，左伍衡1 日期：2009-12-30/span> 浏览：3901 查看PDF文档

纹理图像的相位场演化分割研究*

周佳男1，2，刘均1，冯志林1，赵端阳1，左伍衡1
(1.浙江工业大学 信息学院，浙江 杭州 310014；2.浙江商业职业技术学院 信息技术学院，浙江 杭州 310053)

摘要：为研究在噪声环境下的纹理图像的分割,提出了一种基于相位场演化的纹理图像分割算法。首先在相位场模型中加入了形状统计的能量项,以此指导和约束边缘界面轮廓的演化过程;然后通过对改进相位场模型的梯度流方程的求解，得到了算法的数值解。最后将本研究算法用于纹理图像的分割。实验结果验证了该算法的有效性。
关键词：相位场；图像分割；形状特征
中图分类号：TP317.4文献标识码：A文章编号：1001-4551(2009)11-0062-03

Texture image segmentation algorithm based on phase field evolution
ZHOU Jianan1,2, LIU Jun1, FENG Zhilin1, ZHAO Duanyang1, ZUO Wuheng1
(1. College of Information Engineering, Zhejiang University of Technology, Hangzhou 310014, China;
2. College of Information Engineering, Zhejiang Vocational College of Commerce, Hangzhou 310053, China)
Abstract: For the study of texture image segmentation under the noisy environments, a novel phase field evolution segmentation algorithm for texture images was proposed. Statistical shape energy was firstly added to phase field model, and could be used to guide and limit the edge interface curve to the target. Then, a numerical solving of the modified model was implemented by its gradient descent flow. Finally, the proposed algorithm had been evaluated on real texture images. The results show the desirable segmentation performance.
Key words: phase field; image segmentation; shape feature
参考文献(References):
［1］岳爱菊,汪西莉.一种变换域纹理图像特征提取和分割方法［J］.计算机工程与应用,2007,43(31):50-54.
［2］夏敏磊,郭斌.基于多级小波分解的两类纹理图像分割技术［J］.机电工程,2007,24(9):115-116.
［3］DU Q, LIU C, WANG X Q. Retrieving topological information for phase field models［J］. SIAM Journal on Applied Mathematics,2005,65(6):1913-1932.
［4］GAJEWSKI H, GARTNER K, GRIEPENTROG J A. On a nonlocal model of image segmentation［J］. Weierstrass Institute for Applied Analysis and Stochastics,2005,56(4):87-107.
［5］DAVID A, ALESSANDRO T. Colour image segmentation by the vectorvalued allencahn phasefield model: a multigrid solution［J］. Numerical Analysis,2007,10(3):180-190.
［6］CREMESR D, TISEHHUASE F, WEIEKERT J, et al. Snakes: introducing statistical shape knowledge into the mumfordshah functional［J］. International Journal of Computer Vision,2002,50(3):295-313.