Range Image Segmentation Based on Differential Geometry: A Hybrid Approach
IEEE Transactions on Pattern Analysis and Machine Intelligence
A Method for Registration of 3-D Shapes
IEEE Transactions on Pattern Analysis and Machine Intelligence - Special issue on interpretation of 3-D scenes—part II
Surface reconstruction from unorganized points
SIGGRAPH '92 Proceedings of the 19th annual conference on Computer graphics and interactive techniques
Edge detection in range images based on scan line approximation
Computer Vision and Image Understanding
Real-time 3D model acquisition
Proceedings of the 29th annual conference on Computer graphics and interactive techniques
Computer Vision
Geometry and texture recovery of scenes of large scale
Computer Vision and Image Understanding
Fast global registration of 3D sampled surfaces using a multi-z-buffer technique
NRC '97 Proceedings of the International Conference on Recent Advances in 3-D Digital Imaging and Modeling
Registration and integration of textured 3-D data
NRC '97 Proceedings of the International Conference on Recent Advances in 3-D Digital Imaging and Modeling
Matching of 3-D curves using semi-differential invariants
ICCV '95 Proceedings of the Fifth International Conference on Computer Vision
Multiview registration for large data sets
3DIM'99 Proceedings of the 2nd international conference on 3-D digital imaging and modeling
Automatic registration and calibration for efficient surface light field acquisition
VAST'06 Proceedings of the 7th International conference on Virtual Reality, Archaeology and Intelligent Cultural Heritage
Hi-index | 0.00 |
We are developing a system that can synchronistically acquire laser range data and visual image data to build reality 3D models of urban environments. A major bottleneck in the process is automatically aligning all range scans into a common corrdinate system. In this paper, we propose a new method for automatic registration of multiple 3D data sets using feature units. Our algorithm first extracts 3D features from range scans, defines virtual features, and constructs feature units. Virtual features complement some missing information due to occlusions and feature units describe the relationship between features so that our method is more suitable for large-scale registration. Then, our approach automatically computes local registrations between individual scans with feature units, builds a topological graph using minima spanning tree (MST) and places all scans in the same coordinate system. Experiment results verify that our algorithm is quite effective in the construction of the urban scenes even with heavy occlusion.