Current Issue Cover
面向RGBD深度数据的快速点云配准方法 (NCIG2016)


摘 要
A method of fast point cloud registration based on RGBD data

Su Benyue,Ma Jinyu,Peng Yusheng,Sheng Min,Ma Zuchang()

Objective: Three-dimensional reconstruction of real objects has always been a hot topic in computer graphics, computer vision and other fields. In response to the 3D reconstruction of object which is obtained with non-uniform rotating from non-fixed angle, a registration method of reconstruction object 3D model based on RGBD data is presented by using a rotating platform. Method: Firstly, we use point cloud data of calibration in different angles to calibrate the center axis of the rotating platform, thereby obtain the relative relationship between the Kinect and rotating platform; Secondly, we use the curvature feature of the target point cloud to extract the feature points and find the corresponding points of adjacent point cloud, then exclude the feature points in non-overlapping regions; Thirdly, the dichotomy iterations method is introduced to find the optimal rotation angle around a central axis which meets the minimum of the registration error between the two points cloud. Finally, take the point cloud data that get from any angle to obtain registration under the unified coordinate system and rebuild the model. Results: All experimental results from the common dataset and our own dataset show that this method possesses obvious advantages in the efficiency and accuracy compared with ICP and improved ICP algorithm. Meanwhile, this method is better on texture detail preservation. Conclusions: This paper presents a registration method for the three-dimensional reconstruction with RGBD data. This method simply can realize three-dimensional modeling with the non-uniform rotation from non-fixed angle of the object by single Kinect, it is convenient and practical, suitable for the simple and rapid 3D reconstruction applications.