The contributions of figure and ground textures to segmentation.

Several models of texture segmentation use spatial gradients in the activity of early filters to locate texture boundaries. The models assume that these filters are identical to those involved in the detection and discrimination of near threshold patterns. The models differ in how activity gradients from different types of filters are combined. We examined this question by measuring the respective contributions of a figure and a ground texture to segmentation. Vertical and horizontal line segments were used to construct two perfectly discriminable textures and these textures were used to construct four types of displays. Each display contained an obliquely oriented figure, but the displays differed in the way this figure was defined. Displays consisted of either (1) a horizontally textured figure on a blank background, (2) a blank figure on a vertically textured background, (3) a horizontally textured figure on a vertically textured background or (4) a figure with a mixed texture (50% vertical lines, 50% horizontal lines) on a blank background. In a two-alternative forced-choice experiment, observers were asked to judge the figure's orientation (right or left oblique), and the contrast of the textures was varied across trials. The resulting psychometric functions for segmentation were very similar for the four types of displays, suggesting ways in which a simple model of segmentation should be modified.     

Summation of texture segregation across orientation and spatial frequency: electrophysiological and psychophysical findings

Objects are usually segregated from ground by several visual dimensions. We studied texture segregation in checkerboards defined by gradients in spatial frequency, orientation or both frequency and orientation, using Gabor-filtered noise patterns. Saliency was measured electrophysiologically using the visual evoked potential (VEP) associated with texture segregation ('tsVEP') (an associated component in the visual evoked potential), and psychophysically by a 2AFC task. Spatial frequency and orientation stimuli evoked percepts of texture segregation and tsVEPs in all 11 subjects. The tsVEPs to combined stimuli were larger than those to each dimension alone, but smaller (74%) than the algebraic sum of tsVEPs to both individual dimensions. Psychophysical detection rates differed significantly between all conditions (P < 0.001), with highest rates for the combined stimuli. The findings suggest that segregation based on a combination of 'orientation' and 'spatial frequency' is more salient than that based on either of these alone. The significant deviation from full additivity in the tsVEPs suggests that simultaneous contrasts in spatial frequency and orientation have a common processing stage.     

Temporal resolution of orientation-defined texture segregation: a VEP study

Orientation is one of the visual dimensions that subserve figure-ground discrimination. A spatial gradient in orientation leads to "texture segregation", which is thought to be concurrent parallel processing across the visual field, without scanning. In the visual-evoked potential (VEP) a component can be isolated which is related to texture segregation ("tsVEP"). Our objective was to evaluate the temporal frequency dependence of the tsVEP to compare processing speed of low-level features (e.g., orientation, using the VEP, here denoted llVEP) with texture segregation because of a recent literature controversy in that regard. Visual-evoked potentials (VEPs) were recorded in seven normal adults. Oriented line segments of 0.1 degrees x 0.8 degrees at 100% contrast were presented in four different arrangements: either oriented in parallel for two homogeneous stimuli (from which were obtained the low-level VEP (llVEP)) or with a 90 degrees orientation gradient for two textured ones (from which were obtained the texture VEP). The orientation texture condition was presented at eight different temporal frequencies ranging from 7.5 to 45 Hz. Fourier analysis was used to isolate low-level components at the pattern-change frequency and texture-segregation components at half that frequency. For all subjects, there was lower high-cutoff frequency for tsVEP than for llVEPs, on average 12 Hz vs. 17 Hz (P = 0.017). The results suggest that the processing of feature gradients to extract texture segregation requires additional processing time, resulting in a lower fusion frequency.     

Texture segregation in the cat: a parametric study.

We have investigated how different texture parameters affect texture segregation in the cat, and which strategies cats use to solve the segregation task. Five cats were presented with stimuli consisting of two adjacent panels. One side contained a square area of a particular texture embedded in a different background texture; the other side was filled with only the background texture. The animal's task was to detect at which side the texture difference was presented. Sensitivity for the texture difference was assessed by making one aspect of the texture (in most instances the size of the texture elements) dependent upon performance by means of a staircase procedure. Among the most prominent parametric effects are those of density and element position randomization. In general, segregation was optimal at intermediate densities and deteriorated at larger and smaller densities. Element position randomization caused a slight but systematic decrease in segregation performance. Furthermore, we found texture elements at the border between different textures to be of primary importance for segregation. Which strategy the animals used for solving the segregation task depended upon the presence of random figure/background reversals in subsequent stimulus presentations during training. The animals learned to detect texture differences if these reversals were present, and without reversals, they learned to identify the particular texture in the target square. Interestingly, parameter dependencies of segregation did not depend upon the detection strategy used. We have speculated that the two different strategies used by the cats to solve the segregation tasks are related to different hierarchical levels of texture segregation which can be traced back to different stages of texture processing in human models of segregation performance.     

Image segregation in strabismic amblyopia

Humans with naturally occurring amblyopia show deficits thought to involve mechanisms downstream of V1. These include excessive crowding, abnormal global image processing, spatial sampling and symmetry detection and undercounting. Several recent studies suggest that humans with naturally occurring amblyopia show deficits in global image segregation. The current experiments were designed to study figure-ground segregation in amblyopic observers with documented deficits in crowding, symmetry detection, spatial sampling and counting, using similar stimuli. Observers had to discriminate the orientation of a figure (an "E"-like pattern made up of 17 horizontal Gabor patches), embedded in a 7x7 array of Gabor patches. When the 32 "background" patches are vertical, the "E" pops-out, due to segregation by orientation and performance is perfect; however, if the background patches are all, or mostly horizontal, the "E" is camouflaged, and performance is random. Using a method of constant stimuli, we varied the number of "background" patches that were vertical and measured the probability of correct discrimination of the global orientation of the E (up/down/left/right). Surprisingly, amblyopes who showed strong crowding and deficits in symmetry detection and counting, perform normally or very nearly so in this segregation task. I therefore conclude that these deficits are not a consequence of abnormal segregation of figure from background.     

Attention and visual texture segregation

Visual texture segregation is believed to be performed preattentively. Recent evidence, however, suggests that attention does play an important role. Using visual evoked potentials (VEPs), we investigated the effect of different tasks on texture segregation. Stimuli consisted of Gabor-filtered binary noise patterns. In segregated stimuli, local texture orientation contrasts defined global checkerboard patterns. VEP responses specific to texture segregation were obtained by computing the difference between VEPs to homogeneous and segregated stimuli. Four conditions were examined that required attending either the global pattern, the local structure, random numbers displayed on the screen, or a series of tones. Responses specific to texture segregation were dominated by two occipital negativities peaking around 110 and 230 ms. The earlier one was not affected by the task, whereas the later one was completely abolished when the subjects attended to either numbers or tones (p = .0005 and p = .006, respectively). The results suggest that early stages of texture segregation are not affected by attention, whereas task relevance is crucial for later processes. The timing is compatible with a recurrent processing pattern with initial bottom-up processing of basic stimulus characteristics and a subsequent top-down flow of higher level modulatory information. As attention effects occur across modalities, they cannot be simply explained by competition within the visual cortex.     

A comparison of the dynamics of simple (Fourier) and complex (non-Fourier) mechanisms in texture segregation.

Models of texture segregation frequently feature two processing mechanisms: simple, linear channels (1st-order, Fourier mechanisms) and complex channels (2nd-order, non-Fourier mechanisms). Using texture patterns designed to segregate primarily as a result of activity in one set of channels or the other, we employed the method of cued response to obtain speed-accuracy tradeoff (SAT) functions measuring the time course of texture segregation processing in simple and complex channels. Here, both simple-channel and complex-channel patterns are composed of Gabor-patch texture elements, thus equating input to simple channels and the first stage of complex channels. Subjects were required to identify the orientation of a rectangular texture-region embedded in a background field of a different texture. SAT functions were obtained by requiring subjects to respond within 200 ms after an auditory cue. We found that: (1) when segregation depended primarily on activity in simple channels, performance was faster and better than when it depended primarily on complex channels; (2) in contrast to a previous study (Sutter, A., & Graham, N. (1995). Investigating simple and complex mechanisms in texture segragation using the speed-accuracy tradeoff method. Vision Research, 35, 2825-2843), simple-channel (Fourier) patterns composed of two textured regions were just as easily segregated as simple-channel patterns in which one of the regions was blank instead of textured; (3) performance with complex-channel patterns composed of diagonally oriented Gabor-patches was considerably worse than performance with complex-channel patterns composed of vertically and/or horizontally oriented Gabor-patches; (4) among simple-channel patterns containing only one region of texture (background-only or rectangle-only), there were minimal differences in performance; and (5) as in previous experiments, there were large individual differences in the segregation of complex-channel (non-Fourier) patterns. All of the above results can be explained within the framework of the simple- and complex-channels model of texture segregation.     

Feature-specific electrophysiological correlates of texture segregation

Discrimination between a figure and its surround is an important first step of pattern recognition. This discrimination usually relies, as a first step, on the detection of borders between a figure and its surround, for example based on spatial gradients in luminance, colour, or texture. There is evidence that neurones in the visual cortex are specifically activated by segregation between textures, but the relation between segregation based on different types of features such as colour, luminance, and motion is unclear. Evoked EEG potentials specific to texture segregation were investigated in 17 observers in two separate experiments and by means of functional magnetic resonance imaging in a separate study (Fahle et al., in preparation). Differences in either luminance, colour, line orientation, motion, or stereoscopic depth defined a checkerboard pattern. Patterns defined by each of these features elicited segregation-specific potentials. In contrast to earlier reports (Vision Research 37 (1997) 1409), however, we find pronounced differences between the segregation-specific potentials evoked through different features, especially regarding their peak latencies. The topographical distribution of the activity evoked reveals different polarities and partly specific locations for different stimulus features, indicating the existence of different processors for texture segregation based on different features.     

Different effects from spatial frequency masking in texture segregation and texton detection tasks

The paper reports psychophysical experiments set up to study the visual cues used in texture discrimination. In particular, the special role of "textons", i.e. distinct visual features such as blobs of different size, lines at different orientation, line intersections ("crossings") and line ends ("terminators") which have been proposed to provide the basis of perceptual segregation of texture areas, has been investigated. Texture pairs were briefly presented and simultaneously masked with two-dimensional visual noise at various spatial frequency bands. In different tasks on similar patterns, observers had to estimate the orientation of globally dissecting texture areas ("texture segregation") and to identify and distinguish the texture elements themselves ("texton detection"). Differential masking effects between these tasks indicate that texture segregation is often based on visual cues different from the supposed texton features. The segregation of crossing or terminator differences is also achieved from associated differences in the spatial frequency composition, that of differences in blob size from associated differences in mean luminance. Only differences in line orientation revealed similar masking curves in texture segregation and texton detection tasks.     

Electrophysiological correlates of texture segregation in the human visual evoked potential.

We investigated whether the visual evoked potential (VEP) reflects cortical processing associated with preattentive texture segregation. On a visual display unit we presented stimuli with various arrangements of oriented line segments that either led to the appearance of a "preattentive" checkerboard or did not. Two presentation modes were used (pattern onset at 1 Hz and rapid pattern change at 4.3 Hz), while luminance (57 cd/m2) and contrast (92%) of the line segments remained constant. VEPs were recorded in 7 human subjects. The VEP was analyzed as a linear combination of putative components, which are evoked by either local pattern, quasi-local orientation contrast or global preattentive structure. In the transient VEP, we found a negativity over the posterior pole at a latency between 161 and 225 msec (FWHM) in the linear combination designed to extract segregation-specific components. Peak amplitude reached 3.1 +/- 0.8 microV (mean +/- SEM) at 199 msec. This negative peak appeared only for textures containing orientation contrast. Steady-state analysis of the rapid presentation also revealed a significant component (P = 0.002) associated with texture segregation. These potentials either represent processing of orientation contrast or global processing of texture segregation. The results suggest that specific surface potentials, differing from cognitive potentials, can be derived which are associated with preattentive processing.     

Some observations on the effects of slant and texture type on slant-from-texture

We measure the performance of five subjects in a two-alternative-forced-choice slant-discrimination task for differently textured planes. As textures we used uniform lattices, randomly displaced lattices, circles (polka dots), Voronoi tessellations, plaids, 1/f noise, "coherent" noise and a leopard skin-like texture. Our results show: (1) Improving performance with larger slants for all textures, (2) and some cases of "non-symmetrical" performance around a particular orientation. (3) For orientations sufficiently slanted, the different textures do not elicit major differences in performance, (4) while for orientations closer to the vertical plane there are marked differences among them. (5) These differences allow a rank-order of textures to be formed according to their "helpfulness"--that is, how easy the discrimination task is when a particular texture is mapped on the plane. Polka dots tend to allow the best slant discrimination performance, noise patterns the worst. Two additional experiments were conducted to test the generality of the obtained rank-order. First, the tilt of the planes was rotated by 90 degrees. Second, the task was changed to a slant report task via probe adjustment. The results of both control experiments confirmed the texture rank-order previously obtained. We then test a number of spatial-frequency-based slant-from-texture models and discuss their shortcomings in explaining our rank-order. Finally, we comment on the importance of these results for depth-perception research in general, and in particular the implications our results have for studies of cue combination (sensor fusion) using texture as one of the cues involved.     

Figure–ground discrimination in the avian brain: The nucleus rotundus and its inhibitory complex

In primates, neurons sensitive to figure–ground status are located in striate cortex (area V1) and extrastriate cortex (area V2). Although much is known about the anatomical structure and connectivity of the avian visual pathway, the functional organization of the avian brain remains largely unexplored. To pinpoint the areas associated with figure–ground segregation in the avian brain, we used a radioactively labeled glucose analog to compare differences in glucose uptake after figure–ground, color, and shape discriminations. We also included a control group that received food on a variable-interval schedule, but was not required to learn a visual discrimination. Although the discrimination task depended on group assignment, the stimulus displays were identical for all three experimental groups, ensuring that all animals were exposed to the same visual input. Our analysis concentrated on the primary thalamic nucleus associated with visual processing, the nucleus rotundus (Rt), and two nuclei providing regulatory feedback, the pretectum (PT) and the nucleus subpretectalis/interstitio-pretecto-subpretectalis complex (SP/IPS). We found that figure–ground discrimination was associated with strong and nonlateralized activity of Rt and SP/IPS, whereas color discrimination produced strong and lateralized activation in Rt alone. Shape discrimination was associated with lower activity of Rt than in the control group. Taken together, our results suggest that figure–ground discrimination is associated with Rt and that SP/IPS may be a main source of inhibitory control. Thus, figure–ground segregation in the avian brain may occur earlier than in the primate brain.     

Attention modulates psychophysical and electrophysiological response to visual texture segmentation in humans

To investigate whether processing underlying texture segmentation is limited when texture is not attended, we measured orientation discrimination accuracy and visual evoked potentials (VEPs) while a texture bar was cyclically alternated with a uniform texture, either attended or not. Orientation discrimination was maximum when the bar was explicitly attended, above threshold when implicitly attended, and fell to just chance when unattended, suggesting that orientation discrimination based on grouping of elements along texture boundary requires explicit attention. We analyzed tsVEPs (variations in VEP amplitude obtained by algebraic subtraction of uniform-texture from segmented-texture VEPs) elicited by the texture boundary orientation discrimination task. When texture was unattended, tsVEPs still reflected local texture segregation. We found larger amplitudes of early tsVEP components (N75, P100, N150, N200) when texture boundary was parallel to texture elements, indicating a saliency effect, perhaps at V1 level. This effect was modulated by attention, disappearing when the texture was not attended, a result indicating that attention facilitates grouping by collinearity in the direction of the texture boundary.     

Temporal resolution of orientation-based texture segregation

We analysed the temporal-frequency characteristics of two functional processes involved in orientation-based texture segregation: local orientation coding and subsequent orientation-contrast coding. Two texture images, in which each micropattern was rotated by 90 degrees, were alternated at various temporal frequencies. A micropattern was a second-derivative (D2) of a Gaussian that loses orientation information when temporally fused with the orthogonal D2 pattern. We measured the upper temporal-frequency limits for localising the target region whose mean orientation differed from the background by 90 degrees or by 45 degrees. If the temporal limit of the texture perception is determined by the most sluggish processing stage, the temporal limit for the 90 degrees texture should be determined by local orientation coding or by orientation-contrast coding, depending on which stage has the lower temporal precision. On the other hand, the 45 degrees texture should always be segregated below the temporal limit of local orientation coding regardless of the temporal limit of orientation-contrast coding. We found that the temporal limit for the 90 degrees texture was slightly higher than that for the 45 degrees texture under spatial conditions appropriate for texture segregation. Moreover, an orientation-noise analysis of segregation performance for a wide range of temporal frequencies revealed that the temporal-frequency sensitivities for the two textures were nearly identical. These results imply that the temporal limit for orientation-based texture segregation depends only on that of local orientation coding. This conclusion further suggests that the potential temporal resolution of orientation-contrast coding is not lower than that of local orientation coding, which would imply that the orientation-contrast coding is unlikely to be mediated by sluggish neural processes.     

Discrimination of an orientation difference in dynamic textures

We investigated whether the response of a motion sensor was related to the specificity of sensory information (orientation and direction of motion) used to compute motion energy. This was done in two ways. First, we assessed whether orientation discrimination of a target line, which segregated by an orientation difference from a textured background, was improved with two-frame apparent motion stimulation (as compared with static presentation). Second, we investigated whether the amount of improvement (in either orientation or direction of motion discrimination) depends on a particular combination of target orientation and direction of motion (either orthogonal or parallel). We found that the percentage of correct responses in the discrimination task (a) was higher for a moving target than for a static one; (b) was higher when the target was oriented more orthogonally to motion direction than background elements; (c) was little affected by background motion and (d) decreased with frame duration in the direction of motion task whereas it was largely unaffected by frame duration in the discrimination of orientation task. These results suggest that discrimination of moving texture boundaries is based on a motion sensor tuned to a particular combination of orientation and direction of motion, which is capable of signalling the orientation of a moving target more accurately than a static sensor.     

A new cue to figure-ground coding: top-bottom polarity

We present evidence for a new figure-ground cue: top-bottom polarity. In an explicit reporting task, participants were more likely to interpret stimuli with a wide base and a narrow top as a figure. A similar advantage for wide-based stimuli also occurred in a visual short-term memory task, where the stimuli had ambiguous figure-ground relations. Further support comes from a figural search task. Figural search is a discrimination task in which participants are set to search for a symmetric target in a display with ambiguous figure-ground organization. We show that figural search was easier when stimuli with a top-bottom polarity were placed in an orientation where they had a wide base and a narrow top, relative to when this orientation was inverted. This polarity effect was present when participants were set to use color to parse figure from ground, and it was magnified when the participants did not have any foreknowledge of the color of the symmetric target. Taken together the results suggest that top-bottom polarity influences figure-ground assignment, with wide base stimuli being preferred as a figure. In addition, the figural search task can serve as a useful procedure to examine figure-ground assignment.     

Combination of texture and color cues in visual segmentation

The visual system can use various cues to segment the visual scene into figure and background. We studied how human observers combine two of these cues, texture and color, in visual segmentation. In our task, the observers identified the orientation of an edge that was defined by a texture difference, a color difference, or both (cue combination). In a fourth condition, both texture and color information were available, but the texture and color edges were not spatially aligned (cue conflict). Performance markedly improved when the edges were defined by two cues, compared to the single-cue conditions. Observers only benefited from the two cues, however, when they were spatially aligned. A simple signal-detection model that incorporates interactions between texture and color processing accounts for the performance in all conditions. In a second experiment, we studied whether the observers are able to ignore a task-irrelevant cue in the segmentation task or whether it interferes with performance. Observers identified the orientation of an edge defined by one cue and were instructed to ignore the other cue. Three types of trial were intermixed: neutral trials, in which the second cue was absent; congruent trials, in which the second cue signaled the same edge as the target cue; and conflict trials, in which the second cue signaled an edge orthogonal to the target cue. Performance improved when the second cue was congruent with the target cue. Performance was impaired when the second cue was in conflict with the target cue, indicating that observers could not discount the second cue. We conclude that texture and color are not processed independently in visual segmentation.     

Within-texture collinearity improves human texture segmentation

Spatial arrangement has been shown to facilitate both detection of a threshold target by collinear flankers and detection of smooth chains within random arrays of suprathreshold elements. Here, we investigate the effect of alignment between texture elements on orientation-based texture segmentation. Textures composed of Gabor elements were used in a figure-discrimination task. The degree of collinearity within the texture was manipulated, and threshold figure-ground orientation differences found. A facilitative effect of collinearity on segmentation was seen, which was insensitive to Gabor carrier phase at the texture-element co-axial spacing of 3lambda used here. The pattern of results with respect to collinearity could not be attributed simply to improved linkage of local orientation contrast at figure borders in isolation, and instead suggests a role for the figure interior in texture segmentation.     

Texture discrimination: representation of orientation and luminance differences in cells of the cat striate cortex

Neuronal texture discrimination in the cat striate cortex was investigated by measuring the responses of single cells to different pattern structures. The representation of two independent features, texture orientation and texture luminance, was analysed in detail and the sensitivity of neurones to either feature was studied at different levels of structure density. Texture patterns were systematically moved across the receptive field. From the cell response to various parts of the pattern, "response patterns" were generated which displayed the cell transform of the textured stimulus pattern. Only when texture structures were coarse, were cells able to encode the texture orientation of an area. Differences in texture luminance, on the other hand, were detected only in fine texture structures. Further, these textural features were processed in a different manner: Cells responded to differences in texture luminance but continuously to areas of similar texture orientation. Thus, responses of striate cells reveal an ambiguous representation of texture features and a failure to uniquely encode texture borders.     

Sensitivity for structure gradient in texture discrimination tasks

Recent experiments indicate that the segregation of visual structures ("texture discrimination") depends not only on the form of texture elements but also on their spacing. Structures with discriminable elements in close proximity can be segregated more easily than patterns in which the same texture elements are more widely spaced. In dot arrays with areas of different dot luminance, segregation was found to depend on both the luminance difference and dot spacing; discrimination of texture areas in coarse dot rasters required greater differences in luminance than in fine rasters. Also, in regular arrays of iso-luminant line patterns, the maximal spacing between neighbouring lines for which different texture areas could still be discriminated was found to be influenced by the degree of dissimilarity between elements. For lines of a given length, texture areas with small differences in orientation became indiscriminable at smaller spacings than texture areas with orthogonal line orientations. Line length additionally had a strong effect on texture discrimination; increasing the line length for a given spacing provided easier segregation of texture areas. However, over a range of raster widths, discrimination of texture areas with a given difference in line orientation varied not with absolute values of line length but with the ratio of line length to interline spacing. Overall, the data suggest that texture discrimination in man is based on the evaluation of variation in structure over space (defined as the "texture gradient"). If local variation of structure is too small, texture areas cannot be discriminated, though differences between texture elements themselves may be apparent. As far as the dependence on variation over space is concerned, discrimination of iso-luminant textures resembles the limited sensitivity of the visual system for differences in texture luminance.