基于机器视觉的工业仪表识别技术研究  

Research on Industrial Analog Instrument Recognition Based on Machine Vision

作　　者：陈梦迟 黄文君[1] 张阳阳 吴元元 李德文 王昊 CHEN Meng-chi;HUANG Wen-Jun;ZHANG Yang-yang;WU Yuan-yuan;LI De-wen;Wang Hao(College of Control Science and Engineering,Zhejiang University,Hangzhou 310027,China;Zhejiang NHU Co.,Ltd.,Shaoxing 312500,China;Zhejiang SUPCON Technology CO.,Ltd.,Hangzhou 310053,China)

机构地区：[1]浙江大学控制科学与工程学院,浙江杭州310027 [2]浙江新和成股份有限公司,浙江绍兴312500 [3]浙江中控技术股份有限公司,浙江杭州310053

出　　处：《控制工程》2020年第11期1995-2001,共7页Control Engineering of China

基　　金：高技术船舶科研计划项目(工信部联装[2016]26号)。

摘　　要：提出了一种基于机器视觉的二维码辅助式工业指针式仪表识别技术方案。本方案将表盘信息编辑成信息帧并制成二维码粘贴在表盘上,在识别图像前先获取待识别仪表的特征信息并将其作为算法重要的参数来源,从而实现不同类型仪表的自适应识别。此外,方案还利用了二维码的定位和校正图案,通过透视变换对图像进行倾斜校正,克服了因拍摄角度倾斜所造成的图像畸变的影响。表盘与指针的定位提取则利用Hough变换实现。实验结果证明,算法在移动终端上具有良好的识别效果及鲁棒性。In this paper,a kind of quick response code(QR code)-assisted industrial instrument recognition solution is proposed.The features of the analogy dial are compiled into certain information frame,which will be converted into QR code and attached to the dial later.Before recognition,the algorithm acquires feature information by scanning QR code firstly,which is the important source of algorithm parameters,so as to realize the recognition of different types of instruments adaptively.Meanwhile,the position and alignment patterns of QR code are also used to correct image to overcome the adverse effect of distortion caused by the tilt shooting angle.Besides,the location and extraction of the dial region and pointer are achieved by Hough transform.The experimental results show that the algorithm proposed in this paper has a good performance of recognition and robustness on mobile terminals.

关 键 词：机器视觉 二维码空间定位 图像校正 HOUGH变换 

分 类 号：TN911.73[电子电信—通信与信息系统]