基于图像处理的钢管字符分割方法研究  

Research on Steel Pipe Character Segmentation Method Based on Image Processing

作　　者：刘红杰[1] 杨亚楠[1] 温佳[1] 韩睿 LIU Hongjie;YANG Yanan;WEN Jia;HAN Rui(School of Electrical and Electronic Engineering,Tiangong University,Tianjin 300072,China;School of Electrical Automation and Information Engineering,Tiangong University,Tianjin 300072,China)

机构地区：[1]天津工业大学电气与信息工程学院,天津300072 [2]天津工业大学电气自动化与信息工程学院,天津300072

出　　处：《实验室研究与探索》2022年第6期138-142,共5页Research and Exploration In Laboratory

基　　金：国家重点研发计划专项(2017YFB1104202)。

摘　　要：钢管字符作为钢管的“身份证”号需清晰记录,但易受氧化、生锈等因素的影响,发生缺损或黏连。为解决受腐蚀字符以及黏连断裂字符的分割问题,设计一种改进的钢管字符分割算法。对原始图像进行二值化等预处理操作,利用黑白像素特点对图像进行修补填充,根据连通域的面积属性去除小噪声,将图像进行分类处理。通过最小二乘法对字符图像进行倾斜矫正,根据连通域边界外接框属性对字符部分进行定位,采用连通域提取法和上下轮廓分析法两种分割算法层层递进的方式完成分割任务。经验证,该方法可以较准确的分割出反光、模糊、断裂和黏连字符图像且明显优于传统方法。The character of the steel pipe is the“ID card”number of the steel pipe and it needs to be properly recorded,but is easily affected by factors such as oxidation and rust,resulting in defects and adhesions.In order to solve the segmentation problem of corroded characters and sticky broken characters,an improved steel pipe character segmentation algorithm is designed.After preprocessing operations such as binarization of the original image,the image is patched and filled with the characteristics of black and white pixels,and the small noise is removed according to the area attribute of the connected domain,and then the image is classified.The character image is tilted by the least square method,and the character part is located according to the bounding box attribute of the connected domain.Finally,two segmentation algorithms,the connected domain extraction method and the upper and lower contour analysis method,are used to complete the segmentation task in a progressive manner.It has been verified that the algorithm can accurately segment the reflective,blurred,broken and hyphenated character images and is significantly better than the traditional algorithm.

关 键 词：图像处理 字符分割 腐蚀 黏连字符 断裂字符 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]