基于图像的数字仪表字轮定位分割算法的研究  

Research of mechanical metercharacter locating and segmenting algorithm based on image processing

作　　者：王臣[1] 严军[1] 朱静[1] 杨其阔 Wang Chen Yan Jun Zhu Jing Yang Qikuo(Communication ＆ Information Department of Shanghai University, Shanghai 200072, Chin)

机构地区：[1]上海大学通信与信息工程学院,上海200072

出　　处：《电子测量技术》2017年第9期132-135,共4页Electronic Measurement Technology

摘　　要：机械式数字仪表的图像识别是一种新兴的抄表技术,该技术通过在仪表前方安装图像采集及处理模组实现。为解决当前图像抄表技术效率低、耗时大等问题,本文提出基于模板匹配的最大稳定极值区域定位分割算法TMMSER(template matching-maximally stable extremal regions)以提高字轮定位分割效率,达到降低设备能耗的目的。TM-MSER算法基于MSER算法,将图像灰度值g和仪表字轮读数区域模板s作为提取参数,将MSER(g)提取算法扩展为二维的TM-MSER(g,s)提取方法。TM-MSER算法在一定程度上排除了背景噪声的干扰,能达到更加快速的定位分割。Mechanical meter image recognition is a kind of new technology in meter reading field, it works by installing image acquisition device in front of meter. In order to solve the problem of low efficiency and high consumption of current technologies, we propose TM-MSER （template matching-maximized stable extrema[ regions） that based on template matching. TM-MSER extends MSER（g） algorithm into two-dimension algorithm, TM-MSER （g, s）, which using gray value and special template as parameters. TM-MSER is more rapid by excluding the background noise interference.

关 键 词：图像处理 TM-MSER算法 定位 分割 

分 类 号：TN919.8[电子电信—通信与信息系统]