An Evolutionary Optimization Method for Parameter Search in 3D Points Cloud Reconstruction

Reconstruction of 3D laser scanned point clouds may generate a mesh characterized by a high number of triangles. Unfortunately, in Computer Aided Design environments neither a simple triangle reduction, nor decimation filters are feasible for mesh optimization, because of their intrinsic errors. In this paper we show how Genocop III can be effectively used to reconstruct a point cloud bounding the error under a certain threshold. Moreover, we define an optimized algorithm for evaluating the reconstruction error, that exploits AABB-trees and pre-computation and provides a useful metric to the genetic algorithm.