中文摘要：

采用KinectFusion点云融合技术,探索三维重建技术与3D打印技术的结合性,设计并实现了一种面向3D打印的点云快速重建算法.首先使用手持型Kinect获取物体表面点云数据,使用八叉树存储数据,利用ICP（iterative closest point）算法进行点云配准与融合;然后采用基于统计异常值检测方法、随机抽样一致性算法（RANSAC）、移动最小二乘法等算法对点云数据进行后处理;再将处理后的点云数据进行三维表面重构并根据重心加入底座、支柱等缺失部位,以保持模型的平稳性;最后使用自制的三角洲打印机打印成型.试验结果表明,该算法实现了从实物到三维虚拟模型再到实物打印成型的整个过程,具有设备成本低、实现简单并且高效快速等特点.

英文摘要：

Using KinectFusion point cloud fusion technology, the combination of 3D reconstruction and 3D printing is explored and a new point cloud reconstruction rapid method for 3D printing is designed and implemented. Surface point cloud data captured by handheld Kinect is stored by octree and ICP （iterative closest point） algorithm is adopted to implement point cloud registration and fusion. The statistical outlier detection method, random sampling consensus algorithm （RANSAC） and moving least squares method are employed to process point cloud data. Then the point cloud data is used in the surface reconstruction step. A base, pillars and other missing parts are added into the model to maintain the stability of the model. Finally the virtual model is printed using homemade Delta printer. Experimental results show that the algorithm implements the process of getting 3D virtual model from real objects and using virtual model for rapid prototyping. The devices are at low price and simple to deploy and the algorithm is effective.