基于改进几何矩的移动机器人目标位姿识别  

Target Pose Recognition of Mobile Robot Based on Improved Geometric Moment

作　　者：朱颖 黄宇钧 张亚婉 唐艳凤 屈福康 ZHU Ying;HUANG Yujun;ZHANG Yawan;TANG Yanfeng;QU Fukang(Guangzhou Huali College,Guangzhou 511325,China)

机构地区：[1]广州华立学院,广州511325

出　　处：《计算机测量与控制》2022年第3期239-243,共5页Computer Measurement &Control

基　　金：广东省普通高校青年创新人才项目(2019KQNCX202);广东省教育厅2016年重点培育学科项目(粤教研函[2017]1号)。

摘　　要：机器人作业环境复杂、物料的随机摆放使得目标识别与定位精度低、实时性差,提出改进几何矩的移动机器人目标识别;采用RGB-D相机进行图像采集与深度信息获取;提出了基于HSV的改进自动阈值与形态学相结合的分割算法对目标物料进行识别,根据HSV颜色空间的特点结合Otsu算法对物料目标进行分割,通过高斯滤波与形态学低通滤波器OC-CO对分割后的目标进行滤波降噪和补全处理;提出了Graham与旋转卡壳相结合的算法寻找最小外接矩来获取目标物料的准确位姿;实验结果表明算法具有较高的准确性和鲁棒性。The robot working environment is complex,and the random material placement makes the precision of object recognition and location low,and the real-time performance is poor,therefore an improved geometric moment for the mobile robot target recognition is proposed.RGB-D camera is used for the image acquisition and the depth information acquisition,and based on HSV and morphological segmentation algorithm,an improved automatic threshold is proposed for the object recognition,according to the characteristics of the HSV color space and the Otsu Algorithm,the material object is segmented,and the segmented object is filtered by the gauss filter and the morphological low-pass filter OC-CO.Based on Graham and rotating jam,an algorithm is proposed to find the minimum external moment and obtain the exact position and pose of the target material.Experimental results show that the Algorithm has high accuracy and robustness.

关 键 词：移动机器人 目标识别 几何矩 凸包 形态学 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]