一维对象复杂度的灰度图像分割算法  

Grayscale image segmentation algorithm based on onedimensional object complexity

作　　者：章怡 王海峰 ZHANG Yi;WANG Haifeng(Information Center,Jiangsu University of Technology,Changzhou Jiangsu 213001)

机构地区：[1]江苏理工学院信息中心,江苏常州213001

出　　处：《太赫兹科学与电子信息学报》2020年第6期1058-1064,共7页Journal of Terahertz Science and Electronic Information Technology

基　　金：江苏省常州市科技计划资助项目(CE20165049);江苏省高校自然科学基金资助项目(18KJB520012)。

摘　　要：从图像复杂度的角度,提出一种一维对象复杂度的灰度图像分割算法。用阈值将灰度图像分为背景与目标2类,统计其对应直方图与总像素个数,并计算对象复杂度;依据图像复杂度分割准则算法公式,遍历每一灰度级对应的图像复杂度值,选取图像复杂度值最小对应的灰度值为最佳分割阈值。仿真实验结果表明,与经典Otsu算法、信息最大熵算法和最小交叉熵算法相比,本文算法速度快,稳定性和效率最好,是一种通用有效的图像分割算法。Inspired by the classical segmentation algorithm,this paper proposes a grayscale image segmentation algorithm based on the image complexity.Firstly,the grayscale image is divided into background and target categories by the threshold,the corresponding histogram and total number of pixels are calculated,as well as the complexity of objects.Secondly,according to the image complexity segmentation criterion,the image complexity of each gray level is calculated.Finally,the optimal segmentation threshold is obtained by the minimum value of the object complexity.Compared with the other three classical algorithms,the experimental results show that the proposed image segmentation algorithm is fast,stable and efficient.

关 键 词：对象复杂度 图像分割 OTSU算法 最大熵 最小交叉熵 

分 类 号：TN21[电子电信—物理电子学] TP393[自动化与计算机技术—计算机应用技术]