Shape Correspondence and Segmentation in Point Clouds


Color point clouds from 3D scanners are a representa-tion of real-world geometry and color. However, such scandata are imperfect, containing noise, outliers, and occlu-sions. Noise-free point clouds can be computed by virtu-ally scanning 3D CAD models from a pre-built library, buttheir geometry may differ from real-world objects. We de-scribe a new algorithm to automatically compute dense cor-respondences between point cloud scans of same-type ob-jects, thus making it possible to transfer real-world colorfrom noisy scans (source scan) to noise-free virtual scans(target scan), even in cases where the scan objects differ.The method segments both point clouds into parts and thencomputes part correspondences between them. An itera-tive algorithm applies a set of rigid transformations to thecorresponding parts to determine a dense mapping betweenthem. The dense mapping allows color or other parametertransfers. The resulting point cloud has the geometry of thevirtual scan and the color from the real-world scan, mappedin a semantically consistent manner.


- Rongqi Qiu and Ulrich Neumann. "IPDC: Iterative Part-Based Dense Correspondence between Point Clouds." IEEE Winter Conference on Applications of Computer Vision (WACV), March 7-9, 2016, Lake Placid, NY, USA. [PDF]
- Rongqi Qiu and Ulrich Neumann. "Exemplar-Based 3D Shape Segmentation in Point Clouds." International Conference on 3D Vision (3DV), October 25-28, 2016, Stanford University, CA, USA. [PDF]

Pipe-Run Extraction and Reconstruction from Point Clouds


We present automatic methods to extract and reconstruct industrial site pipe-runs from large-scale point clouds. We observe three key characteristics in this modeling problem, namely, primitives, similarities, and joints. While primitives capture the dominant cylindric shapes, similarities reveal the inter-primitive relations intrinsic to industrial structures because of human design and construction. Statistical analysis over point normals discovers primitive similarities from raw data to guide primitive fitting, increasing robustness to data noise and incompleteness. Finally, joints are automatically detected to close gaps and propagate connectivity information. The resulting model is more than a collection of 3D triangles, as it contains semantic labels for pipes as well as their connectivity.


- Rongqi Qiu, Qian-Yi Zhou and Ulrich Neumann. "Pipe-Run Extraction and Reconstruction from Point Clouds." European Conference on Computer Vision (ECCV), pages 17-30, September 6-12, 2014, Zurich, Switzerland. [PDF][Video]
- Guan Pang, Rongqi Qiu, Jing Huang, Suya You and Ulrich Neumann. "Automatic 3D Industrial Point Cloud Classification and Modeling." SPE Western Regional Meeting, April 27-30, 2015, Garden Grove, CA, USA.
- Guan Pang, Rongqi Qiu, Jing Huang, Suya You and Ulrich Neumann. "Automatic 3D Industrial Point Cloud Modeling and Recognition." The Fourteenth IAPR International Conference on Machine Vision Applications (oral presentation), May 18-22, 2015, Tokyo, Japan. [PDF]

Voxel-Based Building Reconstruction and Visualization


Our major focus in voxel-based building reconstruction method lies in primitive-based method. In particular, after pre-processing the input data, we go through three steps to create geometric models for individual building models, namely, primitive detection, duplicates elimination, and parameter fitting and modeling. In the first step, i.e., primitive detection, detecting contours based on edge detection is not robust when dealing with noisy data. We use the watershed image processing method to improve the contouring detection algorithm.

NURBS Curve and Surface Blending


Curve and surface blending is an important operation in CAD systems, in which a non-uniform rational B-spline (NURBS) has been used as the de facto standard. In local corner blending, two curves intersecting at that corner are first made disjoint, and then the third blending curve is added-in to smoothly join the two curves with G1- or G2-continuity. In this paper we present a study to solve the joint problem based on curve extension. The following nice properties of this extension algorithm are exploited in depth: (1) The parameterization of the original shapes does not change; (2) No additional fragments are created. Various examples are presented to demonstrate that our solution is simple and efficient.


- Liu Y.J, Qiu R.Q, Liang X.H. “NURBS curve blending using extension.” Journal of Zhejiang University SCIENCE A, 10(4): 570-576, 2009, Springer.
- Liu Yongjin; Qiu Rongqi. “Fairing-optimized CAD method for curvature continuous split joint of NURBS space curves.” (Patent Publication Number: CN101482979)
- Liu Yongjin; Zang Yu; Qiu Rongqi; Zhang Wenqi; Jiang Changhao; Hu Shimin. “Method for split joint of space curves of product external form based on extension.” (Patent Publication Number: CN101299278)