Nonblurred regions show priority for gaze direction over spatial blur

The human eye continuously forms images of our 3D environment using a finite and dynamically changing depth of focus. Since different objects in our environment reside at different depth planes, the resulting retinal images consist of both focused and spatially blurred objects concurrently. Here, we wanted to measure what effect such a mixed visual diet may have on the pattern of eye movements. For that, we have constructed composite stimuli, each containing an intact photograph and several progressively blurred versions of it, all arranged in a 3?×?3 square array and presented simultaneously as a single image. We have measured eye movements for 7 such composite stimuli as well as for their corresponding root mean square (RMS) contrast-equated versions to control for any potential contrast variations as a result of the blurring. We have found that when observers are presented with such arrays of blurred and nonblurred images they fixate significantly more frequently on the stimulus regions that had little or no blur at all (p?<?.001). A similar pattern of fixations was found for the RMS contrast-equated versions of the stimuli indicating that the observed distributions of fixations is not simply the result of variations in image contrasts due to spatial blurring. Further analysis revealed that, during each 5 second presentation, the image regions containing little or no spatial blur were fixated first while other regions with larger amounts of blur were fixated later, if fixated at all. The results contribute to the increasing list of stimulus parameters that affect patterns of eye movements during scene perception.     

Shifts in the perceived location of a blurred edge increase with contrast

Perceived brightness is nonlinearly related to luminance. Consequently, any mechanism operating on the (transformed) luminance profile of a blurred edge to detect its location should make errors, and the magnitude of these errors should increase with contrast. The perceived location of a blurred edge was measured at a range of contrasts and a range of blur space constants in a vernier alignment task. It was found that the perceived location of a blurred edge was affected by the contrast and the blur space constant of the edge. At low contrasts, the apparent location of the blurred edge was near the calculated location of the edge, assuming the linear transduction of luminance. At higher contrasts, the perceived location of a blurred edge was shifted toward the dark side of the edge, and the shift increased with contrast.     

Shifts in the perceived location of a blurred edge increase with contrast

Perceived brightness is nonlinearly related to luminance. Consequently, any mechanism operating on the (transformed) luminance profile of a blurred edge to detect its location should make errors, and the magnitude of these errors should increase with contrast. The perceived location of a blurred edge was measured at a range of contrasts and a range of blur space constants in a vernier alignment task. It was found that the perceived location of a blurred edge was affected by the contrast and the blur space constant of the edge. At low contrasts, the apparent location of the blurred edge was near the calculated location of the edge, assuming the linear transduction of luminance. At higher contrasts, the perceived location of a blurred edge was shifted toward the dark side of the edge, and the shift increased with contrast.     

Perceptual assignment of opacity to translucent surfaces: the role of image blur

In constructing the percept of transparency, the visual system must decompose the light intensity at each image location into two components one for the partially transmissivc surface, the other for the underlying surface seen through it. Theories of perceptual transparency have typically assumed that this decomposition is defined quantitatively in terms of the inverse of some physical model (typically, Metelli's 'episcotister model'). In previous work, we demonstrated that the visual system uses Michelson contrast as a critical image variable in assigning transmittance to transparent surfaces not luminance differences as predicted by Metelli's model [F Metelli, 1974 Scientific American 230(4) 90 98]. In this paper, we study the contribution of another variable in determining perceived transmittance, namely, the image blur introduced by the light-scattering properties of translucent surfaces and materials. Experiment 1 demonstrates that increasing the degree of blur in the region of transparency leads to a lowering in perceived transmittance, even if Michelson contrast remains constant in this region. Experiment 2 tests how this addition of blur affects apparent contrast in the absence of perceived transparency. The results demonstrate that, although introducing blur leads to a lowering in apparent contrast, the magnitude of this decrease is relatively small, and not sufficient to explain the decrease in perceived transmittance observed in experiment 1. The visual system thus takes the presence of blur in the region of transparency as an additional image cue in assigning transmittance to partially transmissive surfaces.     

When texture takes precedence over motion in depth perception

Both texture and motion can be strong cues to depth, and estimating slant from texture cues can be considered analogous to calculating slant from motion parallax (Malik and Rosenholtz 1994, report UCB/CSD 93/775, University of California, Berkeley, CA). A series of experiments was conducted to determine the relative weight of texture and motion cues in the perception of planar-surface slant when both texture and motion convey similar information. Stimuli were monocularly viewed images of planar surfaces slanted in depth, defined by texture and motion information that could be varied independently. Slant discrimination biases and thresholds were measured by a method of single-stimuli binary-choice procedure. When the motion and texture cues depicted surfaces of identical slants, it was found that the depth-from-motion information neither reduced slant discrimination thresholds, nor altered slant discrimination bias, compared to texture cues presented alone. When there was a difference in the slant depicted by motion and by texture, perceived slant was determined almost entirely by the texture cue. The regularity of the texture pattern did not affect this weighting. Results are discussed in terms of models of cue combination and previous results with different types of texture and motion information.     

Kinetic occlusion: further studies of the boundary-flow cue

Previous work has demonstrated that human beings employ a processing assumption, the boundary-flow constraint, in perceiving the order of depth at an edge. Subjects perceive depth order of surfaces on the basis of the relative motions of an image boundary and a projected surface texture on either side of the boundary. In the present study, adult subjects viewed computer-generated kinematograms in which boundary-flow information provided the only cue for depth order. The results of Experiments 1 and 2 indicate that common motion between boundary and texture and differential motion between boundary and texture can independently generate the perception of ordered depths of surfaces. In Experiment 3, we examined the interaction of two processes involved in the extraction of depth order from boundary-flow displays: (1) the propagation of foreground and background surfaces from texture to boundary; and (2) the computation of depth order of surfaces on either side of the boundary. The results indicate that while the mechanism that computes depth from boundary-flow information functions reliably when the mean distance between texture and boundary is 8.1(0), surface propagation may be disrupted for distances of this magnitude.     

面孔知觉中特征、结构和整体加工策略的眼动研究

考察面孔知觉中特征加工、结构加工和整体加工三种不同加工方式对应的眼动模式。实验1中, 将以特征信息为主的错乱面孔和以结构信息为主的模糊面孔作为线索刺激, 引发对测试面孔的特征加工和结构加工, 眼动分析表明：特征加工表现为对面孔各特征内更长的凝视时间, 结构加工表现为对面孔各特征间高频的眼跳。实验2采用相同的研究范式, 将完整面孔、轻度错乱面孔和低水平模糊面孔作为线索刺激, 引发对测试面孔除特征加工和结构加工外的另一种加工方式-- 整体加工, 表现为注视点更多地落在测试面孔中央区的鼻子部位以扩大注视范围, 进而把握整张面孔信息。本研究揭示了三种不同面孔加工方式眼动模式的差异。     

Seeing stereoscopic depth from disparity between kinetic edges

Traditionally, it is assumed that stereovision operates only on the positional difference (disparity) between luminance-defined features in the images in the left and the right eye. Here, I show that stereoscopic depth can be seen from disparity between edges created by relative motion of texture elements, and between edges created by correlated flicker of stationary texture elements. Luminance-based stereopsis was impossible since the texture was binocularly uncorrelated. Positional disparity of the centre of revolving patterns was not an efficient depth cue. Stereopsis from the stimuli presented here was possible even without binocular overlap of textured areas. The results provide evidence that positional disparity of kinetic edges, defined by correlated flicker or motion contrast alone, can be used as matching features to recover stereoscopic depth.     

Inaccurate representation of the ground surface beyond a texture boundary

The sequential-surface-integration-process (SSIP) hypothesis was proposed to elucidate how the visual system constructs the ground-surface representation in the intermediate distance range (He et al, 2004 Perception 33 789-806). According to the hypothesis, the SSIP constructs an accurate representation of the near ground surface by using reliable near depth cues. The near ground representation then serves as a template for integrating the adjacent surface patch by using the texture gradient information as the predominant depth cue. By sequentially integrating the surface patches from near to far, the visual system obtains the global ground representation. A critical prediction of the SSIP hypothesis is that, when an abrupt texture-gradient change exists between the near and far ground surfaces, the SSIP can no longer accurately represent the far surface. Consequently, the representation of the far surface will be slanted upward toward the frontoparallel plane (owing to the intrinsic bias of the visual system), and the egocentric distance of a target on the far surface will be underestimated. Our previous findings in the real 3-D environment have shown that observers underestimated the target distance across a texture boundary. Here, we used the virtual-reality system to first test distance judgments with a distance-matching task. We created the texture boundary by having virtual grass- and cobblestone-textured patterns abutting on a flat (horizontal) ground surface in experiment 1, and by placing a brick wall to interrupt the continuous texture gradient of a flat grass surface in experiment 2. In both instances, observers underestimated the target distance across the texture boundary, compared to the homogeneous-texture ground surface (control). Second, we tested the proposal that the far surface beyond the texture boundary is perceived as slanted upward. For this, we used a virtual checkerboard-textured ground surface that was interrupted by a texture boundary. We found that not only was the target distance beyond the texture boundary underestimated relative to the homogeneous-texture condition, but the far surface beyond the texture boundary was also perceived as relatively slanted upward (experiment 3). Altogether, our results confirm the predictions of the SSIP hypothesis.     

Depth stratification in illusory-contour figures on heterogeneous backgrounds is independent of contour clarity and brightness enhancement

Depth stratification in illusory-contour figures was studied by superimposing Kanizsa figures on heterogeneous backgrounds. Gaussian-noise textures were employed in two rating-scale experiments to explore observers' judgments of (i) illusory-contour clarity, (ii) brightness enhancement, and (iii) depth stratification. In experiment 1, depth stratification was found to be stronger in heterogeneous conditions. In experiment 2, texture coarseness was manipulated to determine how depth stratification is affected by linear elements of different sizes in the background, and to relate the variation to contour clarity and brightness enhancement. Results suggest that depth stratification is independent of contour clarity and brightness enhancement. Preliminary novel observations on illusory-contour formation are reported.     

Effective 3-D shape discrimination survives retinal blur

A single experiment evaluated observers’ ability to visually discriminate 3-D object shape, where the 3-D structure was defined by motion, texture, Lambertian shading, and occluding contours. The observers’ vision was degraded to varying degrees by blurring the experimental stimuli, using 2.0-, 2.5-, and 3.0-diopter convex lenses. The lenses reduced the observers’ acuity from ?0.091 LogMAR (in the no-blur conditions) to 0.924 LogMAR (in the conditions with the most blur; 3.0-diopter lenses). This visual degradation, although producing severe reductions in visual acuity, had only small (but significant) effects on the observers’ ability to discriminate 3-D shape. The observers’ shape discrimination performance was facilitated by the objects’ rotation in depth, regardless of the presence or absence of blur. Our results indicate that accurate global shape discrimination survives a considerable amount of retinal blur.     

Depth from binocular half-occlusions in stereoscopic images of natural scenes

Over the past two decades psychophysical experiments have firmly established that binocular half-occlusions are useful sources of information for the human visual system. The existing literature has focused on simplified stimuli that have no additional cues to depth, apart from stereopsis. From this large body of work we can be confident that the visual system is able to exploit binocular half-occlusions to aid depth perception; however, we do not know if this signal has any influence on perception when observers view complex stereoscopic stimuli with multiple sources of depth information. This issue is addressed here with the use of stereoscopic images of natural scenes, some of which have been digitally altered to manipulate a major half-occlusion signal. Our results show that depth-ordering judgments for these relatively complex stimuli are significantly affected by the nature of the half-occlusion signal, but only when highly textured surfaces are viewed. Under such conditions, the replacement of a binocular half-occlusion with background texture slows reaction time relative to performance when the occluded region is consistent with the foreground object. This result is specific to conditions when the depth ordering is correct (ie not reversed) and depends upon the size of the half-occlusion. The influence of the half-occlusion information in the presence of potent depth cues such as perspective, texture gradient, shading, and interposition is convincing evidence that this information plays a significant role in human depth perception.     

Shrinkage in the apparent size of cylindrical objects

A novel illusion in apparent size is reported. We asked observers to estimate the width and depth of vertically oriented elliptic cylinders depicted with texture or luminance gradients (experiment 1), or the height of horizontally oriented elliptic cylinders depicted with binocular disparity (experiment 2). The estimated width or height of cylinders showed systematic shrinkage in the direction of the gradual depth change. The dissimilarity of 2-D appearance amongst our stimuli implies a large variation in spatial-frequency components and brightness contrasts, eliminating the possibility that these parameters contributed to the illusion. Also, the mechanism inappropriately triggered by pictorial depth cues (eg size scaling) may be irrelevant, because the illusion was obtained even when binocular disparity alone specified the shape of the cylinders. The illusion demonstrated here suggests that our visual system may determine the size of 3-D objects by accounting for their depth structures.     

Transient attention degrades perceived apparent motion

Transient spatial attention refers to the automatic selection of a location that is driven by the stimulus rather than a voluntary decision. Apparent motion is an illusory motion created by stationary stimuli that are presented successively at different locations. In this study we explored the effects of transient attention on apparent motion. The motion target presentation was preceded by either valid attentional cues that attract attention to the target location in advance (experiments 1-4), neutral cues that do not indicate a location (experiments 1, 3, and 4), or invalid cues that direct attention to a non-target location (experiment 2). Valid attentional cues usually improve performance in various tasks. Here, however, an attentional impairment was found. Observers' ability to discriminate the direction of motion diminished at the cued location. Analogous results were obtained regardless of cue type: singleton cue (experiment 1), central non-informative cue (experiment 2), or abrupt onset cue (experiment 3). Experiment 4 further demonstrated that reversed apparent motion is less likely with attention. This seemingly counterintuitive attentional degradation of perceived apparent motion is consistent with several recent findings, and together they suggest that transient attention facilitates spatial segregation and temporal integration but impairs spatial integration and temporal segregation.     

Extremal edges: a powerful cue to depth perception and figure-ground organization

Extremal edges (EEs) are projections of viewpoint-specific horizons of self-occlusion on smooth convex surfaces. An ecological analysis of viewpoint constraints suggests that an EE surface is likely to be closer to the observer than the non-EE surface on the other side of the edge. In two experiments, one using shading gradients and the other using texture gradients, we demonstrated that EEs operate as strong cues to relative depth perception and figure-ground organization. Image regions with an EE along the shared border were overwhelmingly perceived as closer than either flat or equally convex surfaces without an EE along that border. A further demonstration suggests that EEs are more powerful than classical figure-ground cues, including even the joint effects of small size, convexity, and surroundedness.     

Grounding the figure: Surface attachment influences figure-ground organization

We investigated whether the lower region effect on figure-ground organization (Vecera, Vogel, & Woodman, 2002) would generalize to contextual depth planes in vertical orientations, as is predicted by a theoretical analysis based on the ecological statistics of edges arising from objects that are attached to surfaces of support. Observers viewed left/right ambiguous figure-ground displays that occluded middle sections of four types of contextual inducers: two types of attached, receding, vertical planes (walls) that used linear perspective and/or texture gradients to induce perceived depth and two types of similar trapezoidal control figures that used either uniform color or random texture to reduce or eliminate perceived depth. The results showed a reliable bias toward seeing as “figure” the side of the figure-ground display that was attached to the receding depth plane, but no such bias for the corresponding side in either of the control conditions. The results are interpreted as being consistent with the attachment hypothesis that the lower region cue to figure-ground organization results from ecological biases in edge interpretation that arise when objects are attached to supporting surfaces in the terrestrial gravitational field.     

Does visual texture discrimination precede binocular fusion?

Various stereoscopic demonstrations are presented which indicate that visual texture discrimination is based on processes which occur after, or at the same time as, the binocular combination of images from the two eyes. Monocularly invisible texture regions can become apparent, and monocularly visible regions can be hidden, by the processes of binocular fusion.     

Labeling and family resemblance in the discrimination of polymorphous categories by pigeons

Two experiments examined whether pigeons discriminate polymorphous categories on the basis of a single highly predictive feature or overall similarity. In the first experiment, pigeons were trained to discriminate between categories of photographs of complex real objects. Within these pictures, single features had been manipulated to produce a highly salient texture cue. Either the picture or the texture provided a reliable cue for discrimination during training, but in probe tests, the picture and texture cues were put into conflict. Some pigeons showed a significant tendency to discriminate on the basis of the picture cue (overall similarity or family resemblance), whereas others appeared to rely on the manipulated texture cue. The second experiment used artificial polymorphous categories in which one dimension of the stimulus provided a completely reliable cue to category membership, whereas three other dimensions provided cues that were individually unreliable but collectively provided a completely reliable basis for discrimination. Most pigeons came under the control of the reliable cue rather than the unreliable cues. A minority, however, came under the control of single dimensions from the unreliable set. We conclude that cue salience can be more important than cue reliability in determining what features will control behavior when multiple cues are available.     

The perception of doubly curved surfaces from anisotropic textures

Abstract— Most existing computational models of the visual perception of three-dimensional shape from texture are based on assumed constraints about how texture is distributed on visible surfaces. The research described in the present article was designed to investigate how violations of these assumptions influence human perception. Observers were presented with images of smoothly curved surfaces depicted with different types of texture, whose distribution of surface markings could be both anisotropic and inhomogeneous. Observers judged the pattern of ordinal depth on each object by marking local maxima and minima along designated scan lines. They also judged the apparent magnitudes of relative depth between designated probe points on the surface. The results revealed a high degree of accuracy and reliability in all conditions, except for a systematic underestimation of the overall magnitude of surface relief. These findings suggest that human perception of three-dimensional shape from texture is much more robust than would be reasonable to expect based on current computational models of this phenomenon.     

The interaction of oculomotor cues and stimulus size in stereoscopic death constancy.

In the natural world, observers perceive an object to have a relatively fixed size and depth over a wide range of distances. Retinal image size and binocular disparity are to some extent scaled with distance to give observers a measure of size constancy. The angle of convergence of the two eyes and their accommodative states are one source of scaling information, but even at close range this must be supplemented by other cues. We have investigated how angular size and oculomotor state interact in the perception of size and depth at different distances. Computer-generated images of planar and stereoscopically simulated 3-D surfaces covered with an irregular blobby texture were viewed on a computer monitor. The monitor rested on a movable sled running on rails within a darkened tunnel. An observer looking into the tunnel could see nothing but the simulated surface so that oculomotor signals provided the major potential cues to the distance of the image. Observers estimated the height of the surface, their distance from it, or the stereoscopically simulated depth within it over viewing distances which ranged from 45 cm to 130 cm. The angular width of the images lay between 2 deg and 10 deg. Estimates of the magnitude of a constant simulated depth dropped with increasing viewing distance when surfaces were of constant angular size. But with surfaces of constant physical size, estimates were more nearly independent of viewing distance. At any one distance, depths appeared to be greater, the smaller the angular size of the image. With most observers, the influence of angular size on perceived depth grew with increasing viewing distance. These findings suggest that there are two components to scaling. One is independent of angular size and related to viewing distance. The second component is related to angular size, and the weighting accorded to it grows with viewing distance. Control experiments indicate that in the tunnel, oculomotor state provides the principal cue to viewing distance. Thus, the contribution of oculomotor signals to depth scaling is gradually supplanted by other cues as viewing distance grows. Binocular estimates of the heights and distances of planar surfaces of different sizes revealed that angular size and viewing distance interact in a similar way to determine perceived size and perceived distance.