Research is now underway to create a vision system for hardwood log inspection using a knowledge-based approach. In this paper, we present a rule-based, 3-D vision system for locating and identifying wood defects using topological, geometric, and statistical attributes. A number of different features can be derived from the 3-D input scenes. These features and evidence functions are used to compute confidence values for object membership in different defect classes. We will illustrate the use of different knowledge sources in a set of independent and concise rules.
Zhu, Dongping; Conners, Richard W.; Schmoldt, Daniel L.; Araman, Philip A. 1991. CT Image Sequence Analysis for Object Recognition - A Rule-Based 3-D Computer Vision System. Proceedings, 1991 IEEE International Conference on Systems, Man, and Cybernetics. pp. 173-178.