基于局部模糊熵与Otsu的图像阈值分割  

Image Thresholding Segmentation Based on Local Fuzzy Entropy and Otsu

作　　者：张雄美[1] 李钊[2] 易昭湘[1] 石虎林[2] 

机构地区：[1]第二炮兵工程学院,陕西西安710025 [2]第二炮兵驻石家庄地区军事代表室,河北石家庄050081

出　　处：《无线电通信技术》2007年第5期38-40,共3页Radio Communications Technology

摘　　要：传统的图像阈值分割方法对噪声和灰度不均匀敏感,且阈值的确定主要依赖于灰度直方图,未考虑图像中像素的空间信息。通过对含噪声图像像素属性的深入分析,提出了一种基于局部模糊熵和Otsu的图像阈值分割方法。局部模糊熵的滤波作用提高了算法的抗噪性,Otsu的自动阈值选取则提高了算法的可靠性。实验结果表明,该方法能自动、有效地选取阈值,分割效果优于传统的阈值分割方法。Traditional image thresholding segmentation methods are sensitive to noises and intensity inhomogeneities of images.They mainly depend on the histogram to determine the threshold but ignore the spatial information.By analyzing properties of pixels in noisy images,a novel thresholding method based on local fuzzy entropy and Otsu was presented here.The filtering ability achieved by local fuzzy entropy improves the ability of the proposed method to deal with noises,and the automatic threshold selection realized by Otsu enhances its reliability.Experiment results demonstrate that the new method can select the threshold automatically and efficiently,and significantly outperforms the conventional thresholding segmentation methods.

关 键 词：局部模糊熵 OTSU 阈值 图像分割 隶属函数 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]