基于SVM的精密指针式仪表自动读数方法  

Auto-reading Method for Precision Pointer Meter Based on Intersect Cortical Model

作　　者：石睿[1] 谢将剑[1] 赵暄[1] 李剑禹 SHI Rui;XIE Jiangjian;ZHAO Xuan;LI Jianyu(School of engineering, Beijing Forestry University, Beijing 100083)

机构地区：[1]北京林业大学工学院,北京100083

出　　处：《现代制造技术与装备》2016年第11期22-24,共3页Modern Manufacturing Technology and Equipment

基　　金：中央高校基本科研业务费专项资金资助(2016ZCQ08)

摘　　要：针对精密指针式仪表读数算法对自动读数的自适应性差问题,提出一种基于SVM的精密指针式仪表自动读数方法。该方法首先对二值化处理后的仪表图像通过面积形态学提取刻度并做极坐标变化,再利用面积形态学特征提取刻度与指针位置,最后利用SVM识别刻度对应的数值进行读数判读。实验表明,自动读数结果与实际数值相比误差均小于0.1%,算法稳定可靠。Aiming at the problem of adaptivity in autoreadingmethod for precision pointer meter, an auto-readingmethod for precision pointer meter based on support vectormachine (SVM) is proposed. Firstly, the proposed approach extractthe scale by area of morphological features of binary meter image,and polar transformation image is derived by information of arcof the scale. Secondly, the scale and index are extracted in polartransformation image using max-area information of binary polartransformation meter image. Finally, the digits of the scale valuesare recognized by SVM, and the meter reading is calculated by therelative position between scale and index. Experimental resultsshow that the relative error between meter auto-readings and actualreadings is less than 0.1% and the method is reliable.

关 键 词：指针式仪表 支持向量机 数字识别 标准差分析 

分 类 号：TH86[机械工程—仪器科学与技术] TP391.41[机械工程—精密仪器及机械]