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

Abstract

This paper proposes several algorithms for selecting viewpoints, based on information quantities, which provide representative views expressing a whole feature of 3D object. By defining a novel information quantity of Fechner type based on Fechner’s law in psychophysics, we introduce shape information quantities depend on an area of face and depend on a length and sharpness of edge line in a polyhedral object. We then define viewpoint information quantities of several types obtained by summing up shape information quantities of the visible surface form a viewpoint. Representative views are obtained from viewpoints at local maximum of the viewpoint information quantity of each type. The face type and the edge type of algorithms are derived that compute viewpoint information quantities obtained from all visible faces and all visible edge lines respectively. Experimental results and estimation on polyhedral objects and triangular mesh representations of curved objects are presented.