基于OTSU处理不均匀光照图像改进算法研究  

Research on the Improved Algorithm of Processing Uneven Illumination Image Based on OTSU

作　　者：刘明兴 刘泽平 李斌 符朝兴[1] 孟含 LIU Mingxing;LIU Zeping;LI Bin;FU Chaoxing;MENG Han(School of Electromechanic Engineering,Qingdao University,Qingdao 266071,China)

机构地区：[1]青岛大学机电工程学院,山东青岛266071

出　　处：《青岛大学学报（工程技术版）》2020年第3期27-32,共6页Journal of Qingdao University(Engineering & Technology Edition)

摘　　要：为解决不均匀光照下图像二值化问题,提出一种基于最大类间方差法(OTSU算法)的改进二值化算法,将图片分为明亮区域和阴暗区域两部分,分别计算两部分最大类间方差对应的阈值,通过分析阴暗区域特征,判断每一像素点位于明亮区域还是阴暗区域,从而确定每一点的阈值。实验结果表明,该算法可以解决OTSU算法处理光照不均匀图像丢失信息问题,可广泛应用于光照不均匀条件下的文本图像二值化处理,针对特殊情况较好,相对于其他算法适用性更强,本算法可通过提高OTSU算法的运算速度,缩减算法的运行时间。该研究提取信息较为完整,可以作为字符识别及缺陷检测等工作的预处理方法,提高识别精度。In order to solve the problem of image binarization under uneven illumination,an improved binary algorithm based on OTSU algorithm is proposed.The image is divided into two parts:bright region and dark region.The threshold corresponding to the maximum variance between the two parts is calculated respectively.By analyzing the characteristics of dark region,each pixel is judged to be in the bright region or dark region,and the threshold of each point is determined.The experimental results show that the algorithm can solve the problem of information loss in OTSU algorithm,and can be widely used in the binary processing of text image under the condition of uneven illumination.It is better for special cases,and more universal than other algorithms.The algorithm in this paper can reduce the running time of the algorithm by improving the operation speed of OTSU algorithm.This research can be used as a preprocessing method for character recognition and defect detection to improve the recognition accuracy.

关 键 词：二值化 OTSU 不均匀光照 阈值分割 最大类间方差 全局阈值 

分 类 号：TP391.413[自动化与计算机技术—计算机应用技术]