Visual brain and visual perception: how does the cortex do perceptual grouping?

A neural network for enhancing boundaries and surfaces in synthetic aperture radar images

“…Grossberg and Mingolla (1985a, 1985b, 1987) developed the BCS model to simulate how the visual system detcets, completes, and sharpens boundary segmentations in response to a variety of stimuli. Our present work adapts a recent development of the BCS model which suggests how the laminar, columnar, and map organization of the visual eortex accomplishes boundary segmentation (Grossberg et a.l., 1997). For the purposes of image processing we have distilled this full biological model clown to its functional essentials.…”

“…Furthermore, the current small-scale processing is so improved that it forms a useful image processing procedure in its own right, at a speed-up of fifteen times compared to the older, multi-scale algorithm. The present algorithm is both closer in structure to known local circuits in the visual cortex (Grossberg, IV!ingolla, and Ross, 1997;Ross, Grossberg, and Mingolla, 1998) than the previous one, and simpler, containing fewer free parameters. Intuitions about its improved performance can best be appreciated by considering the model of visual ·…”

“…This property refers to the disposition of certain cells in visual cortical areas V1 and V2 to fire in the presence of approximately aligned, but spatially S<~parated, image gradients. According to the Grossberg et al (1997) model, cooperative bipole interactions are achieved by excitatory long-range horizontal pathways among cortical pyramidal cells. In this seetion, these intcraetions will be considered only in the cortical area that is most direetly activated by visual inputs, Vl.…”

“…Intuitions about its improved performance can best be appreciated by considering the model of visual · cortical functioning upon which the algorithm is based. The Grossberg et al (1997) model further develops the Boundary Contour System (BCS) model of Grossberg and Mingolla (1985) by suggesting functional reasons for why the visual cortex is organized into layers, how these layers interact via local networks to form functional columns, and how these columns are embedded into cortical maps, to compute eontext-sensitive and coherent groupings of visual forms that preserve their contrast-sensitivity and spatial context-sensitivity. We interchangeably eall this latter property analog coherence or coherent energy.…”

The complementary brain: unifying brain dynamics and modularity

“…For example, when ratio processing and normalization occur during visual perception, they help to control brightness constancy and contrast 15,16 as well as perceptual grouping and attention 53,54,84,85 . At higher levels of cognitive processing, these mechanisms can provide a neural explanation of the 'limited capacity' of cognitive short-term memory 68 .…”