基于改进GA-Otsu与RANSAC回归的指针式仪表识读算法  

Pointer Meter Reading Algorithm Based on Improved GA-Otsu and RANSAC Regression

作　　者：任志玲[1,2] 曹正言 REN Zhiling;CAO Zhengyan(Faculty of Electrical and Control Engineering,Liaoning Technical University,Huludao Liaoning 125105,China;Erdos Research Insititute,Liaoning Technical University,Erdos Neimenggu 017000,China)

机构地区：[1]辽宁工程技术大学电气与控制工程学院,辽宁葫芦岛125105 [2]辽宁工程技术大学鄂尔多斯研究院,内蒙古鄂尔多斯017000

出　　处：《机械设计与研究》2024年第2期208-213,共6页Machine Design And Research

基　　金：国家自然科学基金资助项目(52177047)。

摘　　要：为解决指针式仪表的人工读数效率低、识读精度不高等问题,提出一种基于改进GA-Otsu与RANSAC回归(随机抽样一致性)的指针式仪表识读方法。利用ABF(自适应双边滤波器)对指针式仪表图像进行纹理和噪声滤除,结合Hough梯度法与Mask掩膜法对仪表图像进行表盘提取。基于改进GA-Otsu的图像分割算法得到分离的指针区域,经过形态学处理提取指针细化图。采用RANSAC算法拟合得到指针中心所在直线,计算其角度值,并结合量程信息与角度法完成仪表读数识读。实验结果表明,该算法能有效地分离指针目标与背景,相较改进前识读速度提升了约42.34%、识读平均相对误差小于1.15%,并对不同光照和阴影干扰均有较强的鲁棒性。In order to solve the problems of low manual reading efficiency and accuracy of pointer instruments,a pointer instrument reading method based on improved GA-Otsu and RANSAC regression is proposed.ABF is used to filter the texture and noise of the pointer instrument image,and the dial image is extracted by Hough gradient method and Mask'mask method.The separated pointer region is obtained by image segmentation algorithm based on improved GA-Otsu,and the pointer thinning map is extracted by morphological processing.RANSAC algorithm is used to fit the straight line where the pointer center is located.The angle value is calculated,and the instrument reading recognition is completed by combining the range information and the angle method.The experimental results show that the proposed algorithm can effectively separate the pointer target from the background.Compared with the original algorithm,the recognition speed is improved by 42.34%,the average relative error of recognition is less than 1.15%,and it is robust against different illumination and shadow interference.

关 键 词：指针式仪表 GA-Otsu RANSAC回归 Hough梯度法 形态学处理 角度法 

分 类 号：TH86[机械工程—仪器科学与技术] TP391[机械工程—精密仪器及机械]