Sign-invariant boundary grouping
Year: 2017
Type: preprint
Abstract: The computational segregation of image parts into foreground and background is an important aspect of 3D scene emergence. 2D images composed of spatially disjoint shapes of the same contrast (black or white) with respect to the gray background, and of opposite contrasts along each induced continuous boundary were computed for the studies summarized here. Perceptual figure-ground judgments for these ambiguous displays are determined by the orientation of contrasts only, not by their relative sign, which has important implications for the design and implementation of unsupervised winner-take-all learning algorithms in intelligent machines that learn to see 2D images in 3D.
Source: HAL
Author: Birgitta Dresp
Language: English
Subfield: Cognitive Neuroscience
Field: Neuroscience
Domain: Life Sciences
Open Access status: green
OpenAlex ID: W4298407335
