Approximate medial axis for CAD models

Several research have pointed out the potential use of the medial axis in various geometric modeling applications. The computation of the medial axis for a three dimensional shape often becomes the major bottleneck in these applications. Towards this end, in a recent work, we suggested an efficient algorithm that approximates the medial axis of a shape from a point sample. The input to this algorithm is only the coordinates of the sample points. As a result the approximation quality is limited by the input sample density. However, in geometric applications involving CAD models, the surfaces from which samples need to be derived are known. In this paper we present heuristics to take advantage of this a priori knowledge in our medial axis approximation algorithm. The quality of the approximation achieved by the method is surprisingly high as our experimental results exhibit.