A Computational Model of Feeling of Depth from Monocular 2-D Figures
Kazushi MURAKOSHI, Katsumasa MIKAMI, and Tadashi KURATA
IEICE Transactions on Information and Systems, Pt.2 (Japanese Edition), Vol. J80-D-II, No. 1, pp. 286-294, 1997
When human see Kanizsa's triangle, he can feel three-dimensional
variation with variation of brightness. The neurons which respond to subjective
contours were found, and their characteristic was modeled. But, there is not a
satisfactory model of three-dimensional variation that is feeling of depth. Thus
we study the mechanism of generating feeling of depth from monocular 2-D figure,
especially, in case of subjective figures. We propose a computational model of
generating feeling of depth from occluding information. In this model, we
postulate that interconnections in V2 from interstripes to thick stripe is
important route for generating feeling of depth. As a result, we could calculate
feeling of depth to separate overlapped figures.
key words: feeling of depth, subjective figure, binocular disparity neuron, occluding information