采用不变矩图像匹配与直方图的仪表识别算法  

Instrument Recognition Algorithm Using Invariant Moment Image Matching and Histogram

作　　者：周戈[1] 曾智[2] 

机构地区：[1]重庆青年职业技术学院信息工程系,重庆400712 [2]中国科学院自动化研究所,北京100190

出　　处：《华侨大学学报（自然科学版）》2017年第2期225-228,共4页Journal of Huaqiao University(Natural Science)

基　　金：重庆市科技攻关计划项目(2011CJ0782);重庆市信息产业发展政策研究重点项目(K2011-52)

摘　　要：为了解决当前仪表识别算法在机床仪表状态信号多变、实时性强的环境下状态识别错误的不足,设计一种基于不变矩图像匹配与直方图的在线仪表识别算法.首先,基于不变矩特征,通过图像的重心坐标获取其中心不变矩,设计图像匹配算子,并根据仪表灯的匹配度量值定位当前状态信号位置.然后,根据定位到的信号中心点坐标,提取出状态信号外接最小矩形.最后,依据信号灯颜色的直方图特性,计算出反投影直方图,实现对状态信号颜色的判断.实验结果表明:与当前机床仪表识别技术相比,面对机床仪表状态信号多变的环境时,文中算法具有更高的识别精度.In order to solve the problem of failure to identify the instrument status in current instrument iden-tification algorithm under the condition of the state signal of the machine instrument changing and the strong- real-time performance, an online instrument recognition algorithm based on invariant moment image matching and histogram was designed. Firstly, image matching operator was designed by obtaining its center invariant moments with the condition through the gravity coordinates of image based on invariant moment feature, and the position of the current status signal was located according to the matching measure of the instrument light. Then the minimum rectangle of the state signal was extracted according to the coordinates of the center point of the signal, and finally,the back projection of histogram was calculated according to the characteristic of the color of the signal light obtaining the color judgment of status signal. The experimental results show that： this algorithm had a higher recognition accuracy facing the changing environment of the state signal of the machine tool compared with the current machine instrument identification technology.

关 键 词：机床仪表 识别算法 不变矩 直方图 图像匹配 反投影 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]