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

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。

Abstract:

_{：The environment of robot is complex, and the random placement of materials makes the precision of object recognition and location low, and the real-time performance is poor, Therefore, an improved geometric moment for mobile robot target recognition is proposed.In this paper, RGB-D camera is used for image acquisition and depth information acquisition, and an improved automatic threshold based on HSV and morphological segmentation algorithm are proposed for object recognition, according to the characteristics of HSV color space and Otsu Algorithm, the material object is segmented, and the segmented object is filtered by gauss filter and morphological low-pass filter OC-CO An algorithm based on Graham and rotating jam is proposed to find the minimum external moment to obtain the exact position and pose of the target material. Experimental results show that the Algorithm has high accuracy and robustness.}