Feature Extraction and Classification from Boundary Representation

Abstract

In the paper, an algorithm for explicit feature extraction and classification from boundary representation is presented. It operates in two phases: the topological and the geometrical. While the topological part is just an adaptation of an already known algorithm, the geometrical part represents an original and new solution. In this part, the algorithm manipulates with features filled by material and the empty ones. The algorithm classifies extracted features into eight classes. It successfully and efficiently handles voids, nested features and many cases of mutual feature intersections. The time complexity depends on input data, and never exceeds O(n^2).