基于机器视觉技术和D—S证据理论的数字显示仪表自动读数  

Reading Digital Display Instrument Automatically Based on Machine Vision and D-S Evidence Theory

作　　者：申中鸿[1] 刘群兴[1] 蒋春旭[1] 赵浩之[1] 

机构地区：[1]工业和信息化部电子第五研究所,广东广州510610

出　　处：《信息工程期刊（中英文版）》2013年第6期122-129,共8页Scientific Journal of Information Engineering

基　　金：受国家科技支撑计划项目（201IBAE01814）,广东省科技计划项目（2011A081304005）,国家自然科学基金项目

摘　　要：利用机器视觉技术与D-S证据理论，实现数字显示仪表的自动读数。首先，利用机器视觉中色彩空间选择，闻值二值化等技术对数显仪表图像进行预处理；之后，统计并提取目标区域像素分布特征；最后，利用D-S证据理论，对所提取的目标区域特征进行融合判断，实现数字自动读取。经过大量实验验证，该算法具有较高的识别准确度与良好的实时性，研究改进后可推广应用于实际工业场所的数显仪表读数。The automatic reading of digital display instrument is achieved by means of the application of machine vision technology and D-S evidence theory; in the accomplishment of which color space selection threshold binarization and other machine vision techniques are applied at first to preprocess images of the digital display instrument; then the characteristics of pixel distribution in the targeted areas are extracted and listed; further, integrated and assessed by D-S evidence theory to realize the automatic reading. It is demonstrated by a number of experiments that the algorithm characterized by high accuracy and recognition in real time can be applicable to practice after imorovement.

关 键 词：机器视觉 D-S证据理论 数字显示仪表 自动读数 

分 类 号：TP[自动化与计算机技术]