计算机与现代化

• 数据库与数据挖掘 • 上一篇    下一篇

基于Kinect的旋转刚体三维重建方法

  

  1. 重庆大学计算机学院,重庆400030
  • 收稿日期:2014-03-21 出版日期:2014-05-28 发布日期:2014-05-30
  • 作者简介:韦羽棉(1984-),男,广西河池人,重庆大学计算机学院硕士研究生,研究方向:模式识别,图像处理; 尚赵伟(1968-),男,甘肃庆阳人,副教授,博士,研究方向:模式识别,图像处理。
  • 基金资助:
    国家自然科学基金资助项目(61173130); 重庆市自然科学基金资助项目(CSTC-2010BB2217)

3D Reconstruction Method of Rotational Rigid Object Based on Kinect

  1. College of Computer Science, Chongqing University, Chongqing 400030, China
  • Received:2014-03-21 Online:2014-05-28 Published:2014-05-30

摘要: 针对在非匀速非定轴旋转条件下利用Kinect进行刚体三维重建问题,提出一种改进的基于Kinect传感器的旋转刚体三维重建方法。首先利用Kinect采集深度图像,然后用改进的加权ICP(Iterative Closest Point)算法在非匀速非定轴旋转条件下进行配准,再将各点云变换到同一坐标系下,最后根据所得点云生成三维模型表面,通过GPU(Graphic Processing Unit)编程技术来提高计算速度以满足实际需求。实验结果表明:该方法具有重建效果良好的特点。

关键词: 三维重建, Kinect, 旋转, 点云, 配准, 刚体

Abstract:  Aiming at the problem of using Kinect for rigid object 3D reconstruction under the condition of non-fixed axis and non-uniform rotation, this paper presents an improved rigid object 3D reconstruction method based on Kinect sensor. First, the depth image is acquired by the Kinect depth camera. Then the improved weighted ICP (Iterative Closest Point) algorithm is used for registration under the condition of non-fixed axis and non-uniform rotation, and all the point clouds are transformed to a universal coordinate system. Finally the 3D model surface is created based on the point cloud. The GPU (Graphic Processing Unit) programming technique is used to raise the calculating speed to meet the actual demand. The experimental results show that this method is of the advantages of good reconstruction effect.

Key words: 3D reconstruction, Kinect, rotation, point cloud, registration, rigid object

中图分类号: