Mining Wooden Pillar Features from Point Cloud
Dean Luo, Yanmin Wang · 2009
For huge quantity of point cloud data gotten from ancient buildings, the common algorithms (e.g. the Hough Transform) can do extract relative features but their efficiency and time expense are not acceptable. In order to improve the data process efficiency and reconstruct their 3D models quickly, more effective and more practicable algorithms should be developed. Here introduce a new algorithm for rapid extracting the wooden pillar features of Chinese ancient buildings from their point cloud data, the algorithm has the least human interaction in the procedure of data processing and is more efficient to extract pillars from point cloud data than existing feature extracting algorithms. With this algorithm we mine wooden pillar features by dividing the point cloud into slices firstly, and then get their projective parameters of relative pillar objects from selected slices, next to compare the local projective parameters in adjacent slices, next to combine them to get the global parameters of wooden pillars and at last reconstruct the 3D wooden pillar models based on acquired global parameters.