Recognizing features from engineering drawings without using hidden lines: a framework to link feature recognition and inspection systems

摘要

Many two-dimensional (2D) feature recognition systems have recently been developed to salvage the massive store of engineering knowledge in 2D form and bring the benefit of computer-aided design (CAD)-computer-aided manufacturing (CAM) integration to 2D CAD users. However, the significant amount of effort spent in developing these systems has little use in the related field of 'Automated Visual Inspection' (AVI). This is because there are some differences in the formats of input to feature recognition and AVI systems with respect to hidden feature information. This paper presents a method to extract feature information from 2D drawings without using hidden lines. This is an initial step towards establishing a common input format between AVI and feature recognition systems, and in turn a common framework to link them. The absence of hidden edges in photographic and video images, which are used as input to the AVI systems, is mimicked by not using hidden lines, although they are normally available in any drawing. However, the lack of hidden line information is compensated for by isometric projections, which are of the same format as that of the photographic input to AVI systems. The ability to recognize features from orthographic projections coupled with isometric views but without hidden lines will be directly transferable to AVI systems, which will have a substantial impact on automatic inspection, reverse modelling and manufacturing automation and, in turn, on manufacturing productivity.