基于SIFT算法的电动车充电孔双目视觉识别技术  

Binocular Visual Recognition Technology for Electric Vehicle Charging Holes Based on SIFT Algorithm

作　　者：于明[1] 李雪薇 YU Ming;LI Xue-wei(School of Engineering,Beijing Forestry University)

机构地区：[1]北京林业大学工学院

出　　处：《化工自动化及仪表》2023年第3期316-322,共7页Control and Instruments in Chemical Industry

基　　金：北京市电动汽车充换电工程技术研究中心开放基金资助项目(2021110038000442)。

摘　　要：基于双目视觉理论提出一种适用于不同光强、背景和任意形状充电插孔的高精度充电孔识别与定位方法。采用SIFT特征提取算法生成高斯差分金字塔进行尺度空间构建,应用机器学习最近邻搜索特征匹配算法获取匹配点映射集合,并配合半全局匹配方法进行视差计算得到高精度充电孔匹配和定位结果。通过试验验证了所提算法的有效性,得到了较为理想的匹配与测距结果。Based on binocular vision theory,a high-precision identification and positioning method for the various charging holes which complying with different light intensity and backgrounds was proposed.in which,the SIFT feature extraction algorithm was adopted to generate a differential Gaussian pyramid for scale space construction,and the nearest neighbor search feature matching algorithm which belongs to machine learning was used to obtain the matching point map set.In addition,the disparity calculation was performed with the semi-global matching method to obtain high-precision matching and positioning results of charging hole.The experimental results verify the effectiveness of the algorithm proposed along with ideal matching and distance measurement results.

关 键 词：充电孔识别和定位 SIFT特征提取 最近邻搜索特征匹配 半全局匹配方法 视差计算 

分 类 号：TP274[自动化与计算机技术—检测技术与自动化装置]