Volume intersection algorithm with changing directions of views

How Many 2D Silhouettes Does It Take to Reconstruct a 3D Object?

“…An object-specific, active apnot strictly relevant to our discussion, it is worth noting that proach could be more effective. Some work in this direction VH(O,V) is also the largest object silhouette-equivalent to has been done by Shanmukh and Pujari [9] and Lavakusha O when observed from viewpoints belonging to V. et al [16]. To discuss the efficiency of the choice of the An object has an infinite number of visual hulls, one for silhouettes requires a suitable definition of reconstruction each possible viewing region.…”

“…raises a number of theoretical questions, namely: (a) which objects are exactly reconstructable; Reconstructing 3D shapes from 2D images is an im-(b) which is the closest approximation that can be obportant research area in computer vision. The technique tained for nonreconstructable objects; known as volume intersection (VI) [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] constructs a rep-(c) what can be inferred about the unknown object O resentation of a 3D object O starting from a set of silhoufrom a reconstructed object R; ettes S i of O. With the word silhouette we indicate the (d) how many silhouettes are necessary for the best region of a 2D image of an object O which contains the possible reconstruction; projections of the visible points of O.…”