Appeal No. 1997-4052
Application 08/479,893
A method for generating a model of a 3-D region of an object
comprising image data representing a plurality of points in the 3-
D region from at least three 2-D images of the appearance of the
region from different positions, comprising the steps of:
(a) calculating, for each point in the 3-D region, at least
one first value representing a variation between image data at
locations on the at least three different 2-D images of the 3-D
region at which each point would appear if the point in the 3-D
region were visible in the 2-D image;
(b) calculating, for each point, using a predetermined rule,
a second value from said first value, said second value
representing a probability that if each point lay on a visible
surface in the 3-D region, the first value would result;
(c) assigning a visibility attribute to each point in
accordance with said second value and visible characteristics of
said locations;
(d) determining from the visibility attribute of each point
from step (c), whether each point in the 3-D region is on the
surface of the object; and
(e) using those points from step (d) which have been
identified to be on the surface of the object to build a model of
the 3-D region for viewing images of the region from multiple
positions.
The references relied upon by the examiner as evidence of
obviousness are:
Miller, Jr. et al. (Miller) 5,224,208 Jun. 29, 1993
(filed Mar. 16,
1990)
Liu et al. (Liu), “3D Curved Object Recognition from Multiple 2D
Camera Views”, Computer Vision, Graphics and Image Processing, May
1990, v.50, n.2, pp. 177-187
2
Page: Previous 1 2 3 4 5 6 7 8 Next
Last modified: November 3, 2007