基于Kinect传感器的移动机器人室内三维环境创建

发布时间：2018-10-14 10:35

【摘要】：针对移动服务机器人需要感知室内工作环境问题,提出了一种基于低成本Kinect传感器的三维环境创建实用方法。机器人在移动过程中,Kinect传感器实时采集RGB信息和深度信息,将RGB图像信息映射到深度图像信息中,采用联合双边滤波对深度图像进行预处理,获得质量比较高的点云数据。对采集到的大量点云数据,采用半径滤波器算法对点云进行精简,剔除离群点,减少点云数量,提高匹配速度。点云配准时,采用NDT算法完成初始配准,从而得到图像帧间粗略的转换关系,并运用GICP算法对采集的多视角点云数据进行精确配准,得到拼接的三维点云场景。实验结果表明:本文方法具有更好的重建效果和更高的效率,可以应用于室内场景三维环境创建。
[Abstract]:Aiming at the problem that mobile service robot needs to perceive indoor working environment, a practical method of creating 3D environment based on low-cost Kinect sensor is proposed. In the process of robot moving, Kinect sensor collects RGB information and depth information in real time, maps RGB image information to depth image information, and uses joint bilateral filter to preprocess depth image to obtain high quality point cloud data. For a large number of point cloud data collected, the radius filter algorithm is used to simplify the point cloud, eliminate outliers, reduce the number of point clouds and improve the matching speed. Point cloud registration on time, NDT algorithm to complete the initial registration, so as to obtain a rough transformation between image frames, and the use of GICP algorithm to accurately register the collected multi-view point cloud data, to obtain the stitched 3D point cloud scene. The experimental results show that the proposed method has better reconstruction effect and higher efficiency and can be applied to create 3D environment of indoor scene.
【作者单位】：南昌大学机电学院;堪萨斯大学电子工程与计算机科学系;
【基金】：国家自然科学基金面上项目(61273282) 江西省科技重点研发计划(20161BBE50090) 江西省自然科学基金(20171BAB201013) 江西省高等学校科技落地计划(KJLD13002)
【分类号】：TP242;TP391.41

【相似文献】