一种改进的二维Otsu阈值分割法  

An Improved Two-dimensional Otsu's Thresholding Segmentation Method

作　　者：谭艳丽[1] 赵永强[1] 

机构地区：[1]太原工业学院电子工程系,太原030008

出　　处：《半导体光电》2014年第5期937-940,共4页Semiconductor Optoelectronics

基　　金：国家自然科学基金项目(60677017)

摘　　要：针对传统灰度级-平均灰度级二维直方图区域划分存在的不足,提出了由灰度级-梯度构成改进的二维直方图区域划分方法,并给出了基于改进二维直方图的Otsu阈值法及其快速递推算法。实验结果表明:提出的改进二维Otsu法与传统二维Otsu法相比,抗噪性能好,分割所得的区域内部具有较好的一致性,获得的目标边缘更加清晰,同时减少了运行时间。According to the obvious deficiencies in the regional division of the traditional gray level-average gray level two-dimensional histogram, an improved regional division method of the gray level-gradient two-dimensional histogram is proposed and meanwhile an Otsu＇s threshold segmentation method and its fast recursive algorithms based on improved two-dimensional histogram are given. The experimental results show that this method achieves better anti-noise ability and the regions of segmentation bear a superior consistency, accurate borders and the computing time of the improved Otsu＇s method is less than the traditional one.

关 键 词：图像分割 阈值选取 二维直方图 OTSU法 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]