改进二维直方图区域划分的阈值分割方法研究  

Image thresholding segmentation research based on an improved region division of two-dimensional histogram

作　　者：李淼[1] 杨恢先[1] 张建波[2] 周彤彤[1] 谭正华[2] 

机构地区：[1]湘潭大学材料与光电物理学院,湖南湘潭411105 [2]湘潭大学信息工程学院,湖南湘潭411105

出　　处：《光电子．激光》2013年第7期1426-1433,共8页Journal of Optoelectronics·Laser

基　　金：湖南省教育厅科研(10C1263);湘潭大学科研(11QDZ11)资助项目

摘　　要：针对传统二维Otsu阈值法存在分割精度不够和错分、二维Otsu曲线阈值法不具普遍性,以及二维直方图θ-划分法存在参数θ取值不能同时兼顾分割结果的抗噪性与边缘形状准确性的状况,提出一种适用更广泛的基于改进二维直方图区域划分的叉分Otsu阈值分割方法。采用两条通过阈值向量点与灰度级轴分别成α和β角(β≥α)的直线划分二维直方图,获得更普遍的区域划分。传统直分法可视为本文方法中α=0°、β=90°的特例;本文方法取α=β=135°时,相当于斜分法;取90°<α=β<180°时,相当θ-划分法。改进的区域划分二维直方图叉分法还可以运用到最大熵等基于二维直方图的阈值分割法,应用范围广泛。实验结果表明,二维直方图叉分法不仅分割结果更准确、抗噪声能力更强,而且分割效率高,具有较好的实用价值。In this paper, a more widely suitable Otsu thresholding method is proposed based on cross division of two-dimensional histogram. The region is divided by two lines both via thresholding vector point. The angle between line one and gray level axis is α and the angle between line two and gray level axis is β, then more ordinary regional division can be acquired. Traditional Otsu thresholding method is regarded as a special case of this method when α=0°, β=90°, while Otsu's curve thresholding segmentation method as a special case when α= β=135°, Otsu's θ-division thresholding segmentation method as a special case when 90°<α=β<180°. This improved two-dimensional histogram cross division method also has a wide application in other thresholding methods, such as maximum entropy which is based on two-dimensional histogram. The experiment results show that the proposed method not only achieves more accurate segmentation results and more robust anti-noise performance, but also has efficient and more practical value in applications.

关 键 词：图像分割 阈值 OTSU 二维直方图 叉分法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]