基于等积环形分割的高频元件快速识别算法  

Fast Identification Algorithm of High Frequency Components Based on Ring Segmentation

作　　者：尹可鑫 欧阳[2] 李柏林[2] 林冬婷 YIN Ke-xin;OU Yang;LI Bo-lin;LIN Dong-ting(State Key Laboratory o£Traction Power,Southwest Jiaotong University,Sichuan Chengdu 610031,China;School of Mechanical Engineering,Southwest Jiaotong University,Sichuan Chengdu 610031,China)

机构地区：[1]西南交通大学牵引动力国家重点实验室,四川成都610031 [2]西南交通大学机械工程学院,四川成都610031

出　　处：《机械设计与制造》2022年第12期196-200,206,共6页Machinery Design & Manufacture

基　　金：四川省重大科技专项—跨媒体智能感知与分析(2018GZDZX0031)。

摘　　要：针对高频元件精密加工后纹理复杂、相似度高,导致较难区分的问题,提出了一种基于等积环形分割的快速定位识别算法。首先使用互信息熵进行快速粗定位,然后在等积环形分割的基础上通过像素梯度幅值与像素梯度方向进行有效特征点的筛选及匹配系数(Matching Ratio Result,MRR)的计算,根据得到的匹配系数对元件进行精确定位及识别分类。实验结果表明,与传统的模板匹配方法相比,该方法的定位精准度和时效性有较大的提升,对1350张高频元件图像的识别准确率达到了95.9%。Aiming at the problem of complex texture and high similarity of high frequency components after pre-cision machining,a fast location recognition algorithm based on equal product ring segmentation is proposed.Firstly,the mutual information entropy is used for fast coarse location,and then on the basis of the equal product ring segmentation,the effective feature points are selected by the gradient amplitude and the gradient direction of the pixel,and the matching ratio result(MRR)is calculated.The components are accurately located and identified and classified according to the matching coefficient.The experimental results show that compared with the traditional template matching method,the accuracy and timeliness of this method are greatly improved,and the recognition accuracy of 1350 high-frequency component images is 95.9%.

关 键 词：高频元件 旋转不变性 互信息熵 等积环形分割 匹配系数 识别分类 

分 类 号：TH16[机械工程—机械制造及自动化] TP277[自动化与计算机技术—检测技术与自动化装置] TH17[自动化与计算机技术—控制科学与工程]