Graph-based recognition of morphological features

Automatic recognition of the shapes of objects represented as solid models is very important in design optimization. Object shape also governs ease of manufacture, ease of orientability, field use and all other life-cycle applications. Characteristic attributes of an object shape such as chamfers, protrusions and depressions, play a significant role in process planning, design for manufacture, etc. These attributes are popularly known as morphological features. In this paper, the problem of identifying such morphological features is divided into two phases: the feature extraction phase and the feature classification phase. In feature extraction, the mechanical part is decomposed into its constituent features such as holes, protrusions and depressions based on the connectivity class in the edge-face graph of the part. In feature classification, on the other hand, the extracted features are identified and classified. This paper describes a feature classification scheme based on topological and geometric attributes of a morphological feature. The feature classification scheme outlined in this paper is capable of identifying new features with minimal human interface.