Viewpoint Selection Based on Fechner Type Information Quantities for 3D Objects

Oba,S., Ikai,T., Aoki,S., Yamashita,T., Izumi,M., Fukunaga,K.

Abstract:
This paper proposes some algorithms for selecting viewpoints, based on information quantities, which provide representative views expressing the whole feature of a 3D object. By defining the novel information quantity of Fechner type based on the Fechner's law in the psychology, we introduce shape information quantities of faces and edge lines of a polyhedral object and viewpoint information quantities of several types that gather them from visible surfaces of a viewpoint. Representative views are obtained from viewpoints at local maximum values of the viewpoint information quantity of each type. These information quantities also apply to the polygonal expression of a 3D curved object as a complex polyhedron. The view potential is known in the psychology, which is the set of qualitatively limited and stable views about the object in spite of viewpoint transference. This view potential is constructed by means of viewpoints at some local maximum and local minimum values of the viewpoint information quantity. Experimental results and estimation are presented.