From filters to features:Scale-space analysis of edge and blur coding in human vision

To make vision possible, the visual nervous system must represent the most informative features in the light pattern captured by the eye. Here we use Gaussian scale-space theory to derive a multiscale model for edge analysis and we test it in perceptual experiments. At all scales there are two stages of spatial filtering. An odd-symmetric, Gaussian first derivative filter provides the input to a Gaussian second derivative filter. Crucially, the output at each stage is half-wave rectified before feeding forward to the next. This creates nonlinear channels selectively responsive to one edge polarity while suppressing spurious or "phantom" edges. The two stages have properties analogous to simple and complex cells in the visual cortex. Edges are found as peaks in a scale-space response map that is the output of the second stage. The position and scale of the peak response identify the location and blur of the edge. The model predicts remarkably accurately our results on human perception of edge location and blur for a wide range of luminance profiles, including the surprising finding that blurred edges look sharper when their length is made shorter. The model enhances our understanding of early vision by integrating computational, physiological, and psychophysical approaches. © ARVO.

Third-derivative filters predict edge locations in spatial vision

Edge detection is crucial in visual processing. Previous computational and psychophysical models have often used peaks in the gradient or zero-crossings in the 2nd derivative to signal edges. We tested these approaches using a stimulus that has no such features. Its luminance profile was a triangle wave, blurred by a rectangular function. Subjects marked the position and polarity of perceived edges. For all blur widths tested, observers marked edges at or near 3rd derivative maxima, even though these were not 1st derivative maxima or 2nd derivative zero-crossings, at any scale. These results are predicted by a new nonlinear model based on 3rd derivative filtering. As a critical test, we added a ramp of variable slope to the blurred triangle-wave luminance profile. The ramp has no effect on the (linear) 2nd or higher derivatives, but the nonlinear model predicts a shift from seeing two edges to seeing one edge as the ramp gradient increases. Results of two experiments confirmed such a shift, thus supporting the new model. [Supported by the Engineering and Physical Sciences Research Council].

Depth propagation and surface construction in 3-D vision

Motion discontinuities can signal object boundaries where few or no other cues, such as luminance, colour, or texture, are available. Hence, motion-defined contours are an ecologically important counterpart to luminance contours. We developed a novel motion-defined Gabor stimulus to investigate the nature of neural operators analysing visual motion fields in order to draw parallels with known luminance operators. Luminance-defined Gabors have been successfully used to discern the spatial-extent and spatial-frequency specificity of possible visual contour detectors. We now extend these studies into the motion domain. We define a stimulus using limited-lifetime moving dots whose velocity is described over 2-D space by a Gabor pattern surrounded by randomly moving dots. Participants were asked to determine whether the orientation of the Gabor pattern (and hence of the motion contours) was vertical or horizontal in a 2AFC task, and the proportion of correct responses was recorded. We found that with practice participants became highly proficient at this task, able in certain cases to reach 90% accuracy with only 12 limited-lifetime dots. However, for both practised and novice participants we found that the ability to detect a single boundary saturates with the size of the Gaussian envelope of the Gabor at approximately 5 deg full-width at half-height. At this optimal size we then varied spatial frequency and found the optimum was at the lowest measured spatial frequency (0.1 cycle deg-1 ) and then steadily decreased with higher spatial frequencies, suggesting that motion contour detectors may be specifically tuned to a single, isolated edge.

Mach edges: a key role for 3rd derivative filters in spatial vision

Edges are key points of information in visual scenes. One important class of models supposes that edges correspond to the steepest parts of the luminance profile, implying that they can be found as peaks and troughs in the response of a gradient (first-derivative) filter, or as zero-crossings (ZCs) in the second-derivative. A variety of multi-scale models are based on this idea. We tested this approach by devising a stimulus that has no local peaks of gradient and no ZCs, at any scale. Our stimulus profile is analogous to the classic Mach-band stimulus, but it is the local luminance gradient (not the absolute luminance) that increases as a linear ramp between two plateaux. The luminance profile is a smoothed triangle wave and is obtained by integrating the gradient profile. Subjects used a cursor to mark the position and polarity of perceived edges. For all the ramp-widths tested, observers marked edges at or close to the corner points in the gradient profile, even though these were not gradient maxima. These new Mach edges correspond to peaks and troughs in the third-derivative. They are analogous to Mach bands - light and dark bars are seen where there are no luminance peaks but there are peaks in the second derivative. Here, peaks in the third derivative were seen as light-to-dark edges, troughs as dark-to-light edges. Thus Mach edges are inconsistent with many standard edge detectors, but are nicely predicted by a new model that uses a (nonlinear) third-derivative operator to find edge points.

Bars and edges: what makes a feature for human vision?

There have been two main approaches to feature detection in human and computer vision - luminance-based and energy-based. Bars and edges might arise from peaks of luminance and luminance gradient respectively, or bars and edges might be found at peaks of local energy, where local phases are aligned across spatial frequency. This basic issue of definition is important because it guides more detailed models and interpretations of early vision. Which approach better describes the perceived positions of elements in a 3-element contour-alignment task? We used the class of 1-D images defined by Morrone and Burr in which the amplitude spectrum is that of a (partially blurred) square wave and Fourier components in a given image have a common phase. Observers judged whether the centre element (eg ±458 phase) was to the left or right of the flanking pair (eg 0º phase). Lateral offset of the centre element was varied to find the point of subjective alignment from the fitted psychometric function. This point shifted systematically to the left or right according to the sign of the centre phase, increasing with the degree of blur. These shifts were well predicted by the location of luminance peaks and other derivative-based features, but not by energy peaks which (by design) predicted no shift at all. These results on contour alignment agree well with earlier ones from a more explicit feature-marking task, and strongly suggest that human vision does not use local energy peaks to locate basic first-order features. [Supported by the Wellcome Trust (ref: 056093)]

The primal sketch revisited: locating and representing edges in human vision via Gaussian-derivative filtering

Marr's work offered guidelines on how to investigate vision (the theory - algorithm - implementation distinction), as well as specific proposals on how vision is done. Many of the latter have inevitably been superseded, but the approach was inspirational and remains so. Marr saw the computational study of vision as tightly linked to psychophysics and neurophysiology, but the last twenty years have seen some weakening of that integration. Because feature detection is a key stage in early human vision, we have returned to basic questions about representation of edges at coarse and fine scales. We describe an explicit model in the spirit of the primal sketch, but tightly constrained by psychophysical data. Results from two tasks (location-marking and blur-matching) point strongly to the central role played by second-derivative operators, as proposed by Marr and Hildreth. Edge location and blur are evaluated by finding the location and scale of the Gaussian-derivative `template' that best matches the second-derivative profile (`signature') of the edge. The system is scale-invariant, and accurately predicts blur-matching data for a wide variety of 1-D and 2-D images. By finding the best-fitting scale, it implements a form of local scale selection and circumvents the knotty problem of integrating filter outputs across scales. [Supported by BBSRC and the Wellcome Trust]

Design of the low vision quality-of-life questionnaire (LVQOL) and measuring the outcome of low-vision rehabilitation

PURPOSE: To design and validate a vision-specific quality-of-life assessment tool to be used in a clinical setting to evaluate low-vision rehabilitation strategy and management. METHODS: Previous vision-related questionnaires were assessed by low-vision rehabilitation professionals and patients for relevance and coverage. The 74 items selected were pretested to ensure correct interpretation. One hundred and fifty patients with low vision completed the chosen questions on four occasions to allow the selection of the most appropriate items. The vision-specific quality of life of patients with low vision was compared with that of 70 age-matched and gender-matched patients with normal vision and before and after low-vision rehabilitation in 278 patients. RESULTS: Items that were unreliable, internally inconsistent, redundant, or not relevant were excluded, resulting in the 25-item Low Vision Quality-of-Life Questionnaire (LVQOL). Completion of the LVQOL results in a summed score between 0 (a low quality of life) and 125 (a high quality of life). The LVQOL has a high internal consistency (α = 0.88) and good reliability (0.72). The average LVQOL score for a population with low vision (60.9 ± 25.1) was significantly lower than the average score of those with normal vision (100.3 ± 20.8). Rehabilitation improved the LVQOL score of those with low vision by an average of 6.8 ± 15.6 (17%). CONCLUSIONS: The LVQOL was shown to be an internally consistent, reliable, and fast method for measuring the vision-specific quality of life of the visually impaired in a clinical setting. It is able to quantify the quality of life of those with low vision and is useful in determining the effects of low-vision rehabilitation. Copyright (C) 2000 Elsevier Science Inc.

Area summation in human vision at and above detection threshold

The initial image-processing stages of visual cortex are well suited to a local (patchwise) analysis of the viewed scene. But the world's structures extend over space as textures and surfaces, suggesting the need for spatial integration. Most models of contrast vision fall shy of this process because (i) the weak area summation at detection threshold is attributed to probability summation (PS) and (ii) there is little or no advantage of area well above threshold. Both of these views are challenged here. First, it is shown that results at threshold are consistent with linear summation of contrast following retinal inhomogeneity, spatial filtering, nonlinear contrast transduction and multiple sources of additive Gaussian noise. We suggest that the suprathreshold loss of the area advantage in previous studies is due to a concomitant increase in suppression from the pedestal. To overcome this confound, a novel stimulus class is designed where: (i) the observer operates on a constant retinal area, (ii) the target area is controlled within this summation field, and (iii) the pedestal is fixed in size. Using this arrangement, substantial summation is found along the entire masking function, including the region of facilitation. Our analysis shows that PS and uncertainty cannot account for the results, and that suprathreshold summation of contrast extends over at least seven target cycles of grating. © 2007 The Royal Society.

A profile of low vision services in England the Low Vision Service Model Evaluation (LOVSME) project

In the UK, low vision rehabilitation is delivered by a wide variety of providers with different strategies being used to integrate services from health, social care and the voluntary sector. In order to capture the current diversity of service provision the Low vision Service Model Evaluation (LOVSME) project aimed to profile selected low vision services using published standards for service delivery as a guide. Seven geographically and organizationally varied low-vision services across England were chosen for their diversity and all agreed to participate. A series of questionnaires and follow-up visits were undertaken to obtain a comprehensive description of each service, including the staff workloads and the cost of providing the service. In this paper the strengths of each model of delivery are discussed, and examples of good practice identified. As a result of the project, an Assessment Framework tool has been developed that aims to help other service providers evaluate different aspects of their own service to identify any gaps in existing service provision, and will act as a benchmark for future service development.

School vision screening, ages 5-16 years:The evidence-base for content, provision and efficacy

The optometric profession in the UK has a major role in the detection, assessment and management of ocular anomalies in children between 5 and 16 years of age. The role complements a variety of associated screening services provided across several health care sectors. The review examines the evidence-base for the content, provision and efficacy of these screening services in terms of the prevalence of anomalies such as refractive error, amblyopia, binocular vision and colour vision and considers the consequences of their curtailment. Vision screening must focus on pre-school children if the aim of the screening is to detect and treat conditions that may lead to amblyopia, whereas if the aim is to detect and correct significant refractive errors (not likely to lead to amblyopia) then it would be expedient for the optometric profession to act as the major provider of refractive (and colour vision) screening at 5-6 years of age. Myopia is the refractive error most likely to develop during primary school presenting typically between 8 and 12 years of age, thus screening at entry to secondary school is warranted. Given the inevitable restriction on resources for health care, establishing screening at 5 and 11 years of age, with exclusion of any subsequent screening, is the preferred option. © 2004 The College of Optometrists.

Neuronal convergence in early contrast vision: binocular summation is followed by response nonlinearity and area summation

We assessed summation of contrast across eyes and area at detection threshold ( C t). Stimuli were sine-wave gratings (2.5 c/deg) spatially modulated by cosine- and anticosine-phase raised plaids (0.5 c/deg components oriented at ±45°). When presented dichoptically the signal regions were interdigitated across eyes but produced a smooth continuous grating following their linear binocular sum. The average summation ratio ( C t1/([ C t1+2]) for this stimulus pair was 1.64 (4.3 dB). This was only slightly less than the binocular summation found for the same patch type presented to both eyes, and the area summation found for the two different patch types presented to the same eye. We considered 192 model architectures containing each of the following four elements in all possible orders: (i) linear summation or a MAX operator across eyes, (ii) linear summation or a MAX operator across area, (iii) linear or accelerating contrast transduction, and (iv) additive Gaussian, stochastic noise. Formal equivalences reduced this to 62 different models. The most successful four-element model was: linear summation across eyes followed by nonlinear contrast transduction, linear summation across area, and late noise. Model performance was enhanced when additional nonlinearities were placed before binocular summation and after area summation. The implications for models of probability summation and uncertainty are discussed.

Optic flow in human vision: MEG reveals a foveo-fugal bias in V1, specialization for spiral space in hMSTs, and global motion sensitivity in the IPS

Abstract We recorded MEG responses from 17 participants viewing random-dot patterns simulating global optic flow components (expansion, contraction, rotation, deformation, and translation) and a random motion control condition. Theta-band (3–7 Hz), MEG signal power was greater for expansion than the other optic flow components in a region concentrated along the calcarine sulcus, indicating an ecologically valid, foveo-fugal bias for unidirectional motion sensors in V1. When the responses to the optic flow components were combined, a decrease in MEG beta-band (17–23 Hz) power was found in regions extending beyond the calcarine sulcus to the posterior parietal lobe (inferior to IPS), indicating the importance of structured motion in this region. However, only one cortical area, within or near the V5/hMT+ complex, responded to all three spiral-space components (expansion, contraction, and rotation) and showed no selectivity for global translation or deformation: we term this area hMSTs. This is the first demonstration of an exclusive region for spiral space in the human brain and suggests a functional role better suited to preliminary analysis of ego-motion than surface pose, which would involve deformation. We also observed that the rotation condition activated the cerebellum, suggesting its involvement in visually mediated control of postural adjustment.

Basic vision: an introduction to visual perception

Review of Basic vision: an introduction to visual perception by R Snowden, P Thompson, T Troscianko; Oxford University Press, Oxford, 408 pages, »27.99 paper (US$59.95) ISBN 9780199286706.

Spatial and temporal dependencies of cross-orientation suppression in human vision.

A well-known property of orientation-tuned neurons in the visual cortex is that they are suppressed by the superposition of an orthogonal mask. This phenomenon has been explained in terms of physiological constraints (synaptic depression), engineering solutions for components with poor dynamic range (contrast normalization) and fundamental coding strategies for natural images (redundancy reduction). A common but often tacit assumption is that the suppressive process is equally potent at different spatial and temporal scales of analysis. To determine whether it is so, we measured psychophysical cross-orientation masking (XOM) functions for flickering horizontal Gabor stimuli over wide ranges of spatio-temporal frequency and contrast. We found that orthogonal masks raised contrast detection thresholds substantially at low spatial frequencies and high temporal frequencies (high speeds), and that small and unexpected levels of facilitation were evident elsewhere. The data were well fit by a functional model of contrast gain control, where (i) the weight of suppression increased with the ratio of temporal to spatial frequency and (ii) the weight of facilitatory modulation was the same for all conditions, but outcompeted by suppression at higher contrasts. These results (i) provide new constraints for models of primary visual cortex, (ii) associate XOM and facilitation with the transient magno- and sustained parvostreams, respectively, and (iii) reconcile earlier conflicting psychophysical reports on XOM.