A method for recognising feature interactions and feature components within the interactions

Handling feature interaction is an outstanding issue in the feature recognition approach. This paper presents a method for recognising the presence of feature interactions and for determining the feature components within the interactions. First, two general feature types, namely depression and protrusion features, are identified from B-rep models based on a modified convex hull concept Secondly, the identified features are represented by a modified AAG (Attributes Adjacency Graph) for facilitating their classification into low-level feature components, such as slots, pockets and bosses. Any unrecognisable features that remain after this second step are regarded as interacting features and they are finally recognised as low-level features via a process of virtual face building and volume adding.