PROJECT TITLE :

A local structure and direction-aware optimization approach for three-dimensional tree modeling - 2016

ABSTRACT:

Modeling three-D trees from terrestrial laser scanning (TLS) purpose clouds remains a difficult task for many well-known reasons, as well as their complicated structure and severe occlusions. In order to accurately reconstruct 3-D tree models from TLS purpose clouds that typically suffer from vital occlusions, during this paper, a novel native structure and direction-aware approach is presented to successfully complete missing structures of trees. During this method, we initial extract the coarse tree skeleton from the input purpose cloud, and therefore, the branch dominant direction and the point density of every branch are obtained. By a skeleton-based Laplacian algorithm, the purpose cloud is any shrunk into a skeleton purpose cloud to highlight the branch dominant direction of every branch. For obtaining even additional accurate purpose densities, a dictionary-based mostly algorithm is utilised to be told and reconstruct the local structure. Finally, the branch dominant direction and purpose density are integrated into an iterative optimization method to recover the missing knowledge. Extensive experimental results have shown that the proposed technique is terribly sturdy to incomplete knowledge sets, and it is capable of accurately reconstructing three-D trees, which are partially, or perhaps to a large extent, missing from the input purpose cloud.