Features extraction from point clouds for automated detection of deformations on automotive body parts

This paper proposes an innovative solution to the problem of extracting feature nodes from a 3D model and grouping nearby feature nodes according to the likelihood that they belong to the same feature. The technique is designed specifically with the problem of detecting unwanted deformations on automotive body part in mind, where feature line detection will not always give the best results. Using an octree representation, the multiresolution method is able to analyze the model for features of various scales. It also uses the octree data structure for feature grouping, and provides an alternative to feature line extraction for connecting similar feature nodes. An existing technique is compared to the proposed approach for feature extraction, and results are presented for the feature grouping method using a point cloud of a miniature car model.