A neural model of early vision: contrast, contours, corners and surfaces - contributions toward an integrative architecture of form and brightness perception
FacultiesFakultät für Informatik
LicenseStandard (Fassung vom 03.05.2003)
The thesis is concerned with the functional modeling of information processing in early and mid-level vision. The mechanisms can be subdivided into two systems, a system for the processing of discontinuities (such as contrast, contours and corners), and a complementary system for the representation of homogeneous surface properties such as brightness. For the robust processing of oriented contrast signals, a mechanism of dominating opponent inhibition (DOI) is proposed. The new mechanism results in a significant suppression of responses to noisy regions, largely independent of the noise level. For the processing of contours, a model of recurrent colinear long-range interaction in V1 is proposed. The key properties of the model are excitatory long-range interactions between cells with colinear receptive fields, inhibitory unoriented short-range interactions and modulating feedback. For the robust and reliable representation and detection of junction points we propose a new scheme, where junctions are characterized by high responses at multiple orientations. The orientation responses can be robustly extracted by the recurrent long-range interaction. A ROC analysis shows that the junction detection is improved compared to a purely feedforward computation of contour orientations. Brightness surfaces are reconstructed by a diffusive spreading or filling-in of sparse contrast measurements, which is locally controlled by contour signals. A mechanism of confidence-based filling-in is proposed where a confidence measure ensures a robust selection of sparse contrast signals. The new mechanism generates brightness surfaces which are invariant against size and shape transformations.
Subject HeadingsVisuelle Wahrnehmung [GND]
Visuelles System [GND]
Neural networks (Computer) [LCSH]
Neural networks: Neurobiology [LCSH]