Key points are not available for this paper at this time.
A computer-assisted technique called shape averaging is presented. Shape averaging produces an abstraction of the typical representation from a set of shapes. Since the averaging is assumed to preserve the characteristics of the original shapes, the result is useful in predicting trends in form or extracting stereotypes from a group of related shapes. The technique can be used to create new forms by blending global features of existing unrelated shapes. The syntactic averaging of shapes consisting of 2-D planar polygons or of 3-D objects represented by planar contours is examined. An algorithm is presented to determine the correspondence between polygons defined by arbitrary numbers of vertices. Algorithms to extract the mean, the median, and the mode from the shapes are also introduced. Potential applications of shape averaging in design are illustrated.>
Chen et al. (Sun,) studied this question.