基于多特征信息的水表指针自动识别方法  

Automatic Recognition Method for Water-meter Pointer Based on Multi-characteristic Information

作　　者：祁建 张燕 

机构地区：[1]江苏电力信息技术有限公司,南京210008

出　　处：《计算机与数字工程》2015年第4期676-679,共4页Computer & Digital Engineering

摘　　要：水表的自动判读技术在工业生产中有着广泛的应用,这种技术主要是通过水表指针的自动识别来实现的。针对水表指针识别中存在的问题,提出了基于多特征信息的指针自动识别方法。首先利用色调信息定位红色指针,结合表盘结构特征实现指针信息显著区域的分割提取;然后采用基于色调信息的P-分位数法提取出指针子表盘区域,较好地解决了低对比度、阴影等干扰因素引起的检测精度下降的问题;最后提出了一种改进的模板匹配法,分区实现了指针指向的快速识别。通过对大量不同条件水表图像的检测,实验结果表明该方法识别精度高、耗时低。The technology of automatic indication recognition for water-meter has wide applications in industrial field. It is realized by recognizing water-meter pointer automatically. In view of existing problems, an improved automatic method is presented based on multi-characteristic information. In this paper, red pointer area is positioned using hue information first- ly, and segmented salience region of pointer information combined with structure character of dial. The sub-dial area is ex- tracted by using P-tile algorithm combined with hue which solves low recognition precision problem caused by interferences such as low contrast and shadow. An optimal template matching is proposed, which recognizes pointer direction by sub-re- gion quickly. The experimental results indicate that the proposed method is higher of accuracy, and lower cost than tradition- al methods.

关 键 词：图像处理 读数识别 P-分位数法 

分 类 号：TH71[机械工程—测试计量技术及仪器]