Psychophysical evidence for fast region-based segmentation processes in motion and color.

Abstract

Theories of image segmentation suggest that the human visual system may use two distinct processes to segregate figure from background: a local process that uses local feature contrasts to mark borders of coherent regions and a global process that groups similar features over a larger spatial scale. We performed psychophysical experiments to determine whether and to what extent the global similarity process contributes to image segmentation by motion and color. Our results show that for color, as well as for motion, segmentation occurs first by an integrative process on a coarse spatial scale, demonstrating that for both modalities the global process is faster than one based on local feature contrasts. Segmentation by motion builds up over time, whereas segmentation by color does not, indicating a fundamental difference between the modalities. Our data suggest that segmentation by motion proceeds first via a cooperative linking over space of local motion signals, generating almost immediate perceptual coherence even of physically incoherent signals. This global segmentation process occurs faster than the detection of absolute motion, providing further evidence for the existence of two motion processes with distinct dynamic properties.A fundamental goal of vision is to locate, characterize, and recognize objects. But to determine "what" is "where" the visual system must first determine which parts of the image belong together. This is the problem of image segmentation, central to both human and machine vision. How the brain implements image segmentation is not known, although various physiological mechanisms have been proposed (for a review, see ref. 1).Objects are distinguished not only by feature contrasts at their boundaries with the background but also by the similarity of feature properties within their boundaries. Two types of segmentation process may exist to exploit these two fundamental distinctions: a local edge-based process that marks differences in visual attributes and a global region-based process that finds homogeneous areas by integrating information about attributes over space. Edge detection is fundamental to many machine vision algorithms (2) but rarely flawless on its own in segmenting an image into relevant regions. In natural images, edges are disrupted by noise, occlusions, and interference from other edges. Edges are also ambiguous: luminance edges, for example, may arise from many distinct physical causes. Therefore, edge detection segmentation algorithms typically require fragile and adaptive adjustment of thresholds, iterations on multiple spatial scales, and special line-completion methods (3, 4). Region-based segmentation algorithms typically "grow" regions from seed patches, accreting all those surrounding areas that share the same properties as the seed, the end result being a set of internally homogeneous regions (5). But region-based algorithms also face problems in defining and setting thresholds for homogeneity, particularly in noisy or complex images characterized by m...

Chromatically rich phenomenal percepts

“…In addition, convincing demonstrations have appeared of activation of concepts that characterize global attributes of complex, abstract stimuli on the basis of brief, single-fixation presentations. One such example is found in a paper by Møller and Hurlbert (1996) on region-based segmentation processes. Subjects briefly (17-200 ms) viewed a dark screen containing a 6.5°square array of 1024 small, randomly positioned colored square dots (4 pixels/dot).…”

“…Analysis of the effects of band eccentricity and width indicated that subjects' identification of its location was based on detection of regional differences on a coarse scale, and not on detection of contrast edges. As the authors (Møller & Hurlbert 1996, p. 7425) put it: "Since the local feature contrast between target and background is minimized by the overlapping color distributions, segmentation requires the integration of signals over space."…”

Animal Detection in Natural Images: Effects of Color and Image Database

“…And it is well established that color information plays an important role in early visual processing [31], [32]. Color has been shown to aid image segmentation [33] and object recognition [34]. Image segmentation is typically used to locate objects and boundaries (lines, curves, etc.)…”

“…in images, which help to determine “what” is “where”. Moller and Hurlbert demonstrated an early contribution of region-based processes to segmentation by color [33]. Gegenfurtner and Rieger used a delayed match-to-sample task to test the role of color vision in the recognition of briefly presented images of natural scenes.…”

Suppressive effects on motion discrimination induced by transient flankers are reduced by perceptual learning

“…An invisible form defined by randomly arranged dots against a similar dotted background becomes immediately visible as soon as it moves, by virtue of the common fate of its dots, which all move together with a common speed and direction. Psychophysics has demonstrated the role of motion in reducing suppressive effects in contrast detection (Lawton, 2000(Lawton, , 2008, texture segmentation (Alberti et al, 2010;Casco et al, 2001;Casco et al, 2006;Mller & Hurlbert, 1996), contour binding (Ledgeway & Hess, 2002;Watamaniuk, McKee, & Grzywacz, 1995), and biological motion (Dittrich, Troscianko, Lea, & Morgan, 1996;Johansson, 1973).…”