基于二维直方图的otsu图像分割算法改进  

Improved otsu image segmentation algorithm based on two-dimensional histogram

作　　者：马胜前[1] 张光南[1,2] 杨金龙[1] 佘乾顺[1] 

机构地区：[1]西北师范大学物理与电子工程学院,甘肃兰州730070 [2]宝鸡文理学院计算机科学系,陕西宝鸡721007

出　　处：《西北师范大学学报（自然科学版）》2009年第1期57-61,共5页Journal of Northwest Normal University(Natural Science)

基　　金：甘肃省科技攻关项目(2GS047-A52-002-07)

摘　　要：分析了灰度图像传统的二维最大类间方差otsu法,在此基础上改进了传统的阈值分割方法,提出了新的阈值分割输出函数,并通过压缩二维直方图预搜索的方法,使运算量大大降低.理论分析和实验结果表明,该方法不仅增强了识别能力,而且大大提高了分割效率.Traditional gray-level image segmentation algorithm based on two-dimensional maximum between-cluster variance histogram-otsu is firstly analyzed, and then improves traditional threshold segmentation method. A new threshold segmentation function is given, which can reduce the computation by compressing two-dimensional histogram and pre-searehing. Theoretical analysis and experiments prove that this method not only enhances the identification capability, but also enhances segmentation efficiency greatly.

关 键 词：灰度图象 阈值分割 二维OTSU法 二维直方图压缩 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]