| Although many machining feature recognition approaches have been developed to assist in the data integration of CAD/CAM systems, extracting machining features from the feature intersections is still a problem.;This thesis comes up with a novel direction to look at the feature intersections and presents a medial axis object based approach to realize the template feature extraction from complex pocket features. The corresponding medial axis and medial axis object graph are generated to transfer the original boundary representation problem to the medial axis object representation problem. The medial axis object-based method not only allows the better integration of the feature shapes and generic shapes, but also supports the better integration of the manufacturing related information into the feature extraction. These two merits help improve the feature intersection analysis and realize more reasonable feature extractions.;The related fundamental concepts of the medial axis transform are introduced and a discrete medial axis transform generation algorithm is proposed in this thesis. The algorithm is designed such that it is able to handle multi-connected boundaries with G1 continuity. The medial axis transform provides another shape representation to deal with shape-related problems like feature extraction. |
