Automatic classification of block-shaped parts based on their 2D projections

This paper presents a classification scheme for 3D block-shaped parts. A pa rt is block-shaped if the contours of its orthographic projections are all rectangles, A block-shaped part is classified based on its partitioned view -contours, which are the result of partitioning the contours of its orthogr aphic projections by visible or invisible projected line;segments. The regi ons and their adjacency in a partitioned view-contour are first converted t o a graph, then to a reference tree, and finally to a vector form, with whi ch a back-propagation neural network classifier can be trained and applied. The proposed back-propagation neural network classifier is in a cascaded s tructure and has advantages that each network can be limited to a small siz e and trained independently. Based on the classification results on their p artitioned view-contours, parts are grouped into families that can be in on e of the three levels of similarity. Extensive empirical tests have been pe rformed; the pros and cons of the approach are also investigated. (C) 1999 Elsevier Science Ltd. All rights reserved.