A multilevel approach to edge detection in tessellated point clouds

Reverse Engineering software packages are not able to automatically and robustly detect the sub-regions of a tessellated model; it generally results in a first trial subdivision which subsequently has to be refined, requiring a large amount of human intervention and high computational effort. To avoid these limitations, the authors propose a multilevel local approach for the automatic identification of edge points in polygonized point clouds, exploiting the ability of a heuristic technique to solve the edge detection problem with low computational effort. The approach was applied to several complex-shaped models with sharp and smooth edges, confirming its robustness and efficiency.