二维Otsu拟合线阈值图像分割方法  

Image Segmentation with Two-dimensional Otsu Fitting Line Threshold Method

作　　者：梁义涛 孟亚敏[1,2] 朱玲艳 张猛 李永刚 LIANG Yi-tao;MENG Ya-min;ZHU Ling-yan;ZHANG Meng;LI Yong-gang(Key Laboratory of Grain Information Processing and Control of Ministry of Education,Henan University of Technology,Zhengzhou 450001,China;College of Information Science and Engineering,Henan University of Technology,Zhengzhou 450001,China)

机构地区：[1]河南工业大学粮食信息处理与控制教育部重点实验室,郑州450001 [2]河南工业大学信息科学与工程学院,郑州450001

出　　处：《科学技术与工程》2021年第9期3689-3697,共9页Science Technology and Engineering

基　　金：国家自然科学基金(31171775);河南省教育厅高等学校重点科研项目(17A510008);河南省教育厅科学技术重点项目(18A520025)。

摘　　要：针对已有二维Otsu线阈值法分割方法存在的因误分类而导致的分割质量下降、抗噪性能不足的问题。结合二维Otsu折线阈值算法和曲线拟合方法,提出了二维Otsu拟合线阈值图像分割方法。本文方法是在二维Otsu折线阈值法基础上进行改进。先对二维直方图中边界信息或噪声所属区域的像素点迭代分割,并设定迭代停止条件,以获得多个阈值点,然后引入曲线拟合的方法,将多个阈值点拟合成线阈值,最后以此线阈值作为分割标准实现分割。实验结果表明:利用本文方法对边缘丰富的图像分割具有较好的分割效果,抗噪能力和自适应能力更强,普适性更高。In the existing segmentation with two-dimensional Otsu line threshold method,the segmentation quality could be reduced due to misclassification and the anti-noise performance is insufficient.An improved image segmentation method combining the two-dimensional Otsu polyline threshold algorithm and the curve fitting method was proposed.In the new method,firstly,the pixels of the boundary information or the area to which the noise belongs in the two-dimensional histogram were iteratively divided,and the iteration stop condition was set to obtain multiple threshold points.Secondly,the curve fitting method was introduced to fit the multiple threshold points into a line threshold.Finally,the line threshold was used as the segmentation criterion to achieve the segmentation.The experimental results show that the new method performs well in image segmentation,especially for those with rich edges,showing stronger anti-noise ability and adaptive ability,and greater applicability.

关 键 词：图像分割 阈值分割 迭代 曲线拟合 线阈值 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]