Similarity Assessment Based on Face Alignment Using Attributed Vectors

TLDR: An algorithm that assesses similarity between two parts by explicitly aligning their faces and hence accounts for these face attributes is described, which involves extracting attributed applied vectors from the face information of parts and aligning the two sets of attributed applied vector sets using rigid body transformations.

Abstract: In certain shape similarity assessment applications, type, locations, and orientations of faces in boundary representations plays a key role in determining similarity between two parts. This paper describes an algorithm that assesses similarity between two parts by explicitly aligning their faces and hence accounts for these face attributes. The approach involves extracting attributed applied vectors from the face information of parts and aligning the two sets of attributed applied vectors using rigid body transformations. The algorithm uses a customizable distance function to align attributed applied vectors. The distance between two aligned attributed applied vector sets is used as a measure of similarity between two parts. This paper also presents computational results to illustrate discrimination capability of the algorithm.