Reconstructing geometry models from scattered data is an important task in reverse engineering. An adaptive subdivision surface reconstruction method was proposed to construct complex models rapidly. This method includes several steps: triangulation on scattered data; mesh segmentation and simplification; computing the subdivision depth according to the specified error. The last step is computing mesh control net by fitting subdivision functions and construct subdivision surface adaptively. In order to improve the efficiency of the algorithm, we implemented the reconstruction algorithm on GPU in parallel way and tested the program on several large scale data sets. Our adaptive subdivision method can save storage space and gain high efficiency simultaneously.
Ying Hui Wang and Wei Yong Wu. Adaptive Subdivision Surface Reconstruction for Scattered Data in Reverse Engineering Based on GPU. Advanced Materials Research, Volumes 490 – 495, 2012. [doi: 10.4028/www.scientific.net/AMR.490-495.138]