基于自适应滤波器的水平集算法-红外技术.PDF

第29 卷 第5 期 红 外 技 术 Vol.29 No.5 2007 年5 月 Infrared Technology May 2007 基于自适应滤波器的水平集算法 余瑞星，朱 冰，张 科，吕梅柏 （西北工业大学航天学院, 陕西 西安 710072 ） 摘要：针对传统水平集无法正确分割含乘性噪声图像的缺点，提出了一种基于自适应滤波器的水平 集算法。该算法首先将自适应滤波器引入到速度停止项中，以消除乘性噪声对水平集演化的影响； 其次，通过估算图像同质区域的相对方差值，达到区分边缘区与同质区以及保持边缘的目的；最后， 将该速度停止项带入到水平集演化方程中，对图像进行演化运算。实验结果表明，该方法能够在抑 制乘性噪声的同时较好地对图像进行分割。 关键词：水平集；乘性噪声；自适应滤波器；速度停止项；图像分割 中图分类号：TP391.41 文献标示码：A 文章编号：1001-8891(2007)05-0279-04 A Level Set Method Based on the Adaptive Filter YU Rui-xing, ZHU Bing ，ZHANG Ke ，LV Mei-bo (College of Astronautics, Northwestern Polytechnical University, Xi’an 710072, China) Abstract ：A new level set method based on adaptive filter was proposed to solve the problem of the traditional level set that can not deal with the image with additive noise. First, a new speed stop item based on adaptive filter was presented to make the evolving front insensitive to the multiplicative noise, then evaluated the variance of homogeneous area was proposed to distinguish edge area from homogeneity area and preserved the edge. Last using the decreasing function of the local contrast image as the stop item evolved the level set function to get the final segmentation result. The experimental results showed that the developed algorithm could segment the image well while suppressing the multiplicative noise. Key words ：level set ；multiplicative noise ；adaptive filter ；speed stop item；image segmentation 此以模糊图像所产生的边缘力来作为水平集的演化动 引言 力，经常会出现不能正确分割图像的情况。为此，本文