Neural mechanisms of human texture processing: texture boundary detection and visual search

Texture of various appearances, geometric distortions, spatial frequency content and densities is utilized by the human visual system to segregate items from background and to enable recognition of complex geometric forms. For automatic, or pre-attentive, segmentation of a visual scene, sophisticated analysis and comparison of surface properties over wide areas of the visual field are required. We investigated the neural substrate underlying human texture processing, particularly the computational mechanisms of texture boundary detection. We present a neural network model which uses as building blocks model cortical areas that are bi-directionally linked to implement cycles of feedforward and feedback interaction for signal detection, hypothesis generation and testing within the infero-temporal pathway of form processing. In the spirit of Jake Beck's early investigations our model particularly builds upon two key hypotheses, namely that (i) texture segregation is based on boundary detection, rather than clustering homogeneous items, and (ii) texture boundaries are detected mainly on the basis of larger scenic contexts mediated by higher cortical areas, such as area V4. The latter constraint provides a basis for element grouping in accordance to the Gestalt laws of similarity and good continuation. It is shown through simulations that the model integrates a variety of psychophysical findings on texture processing and provides a link to the underlying physiology. The functional role of feedback processing is demonstrated by context dependent modulation of V1 cell activation, leading to sharply localized detection of texture boundaries. It furthermore explains why pre-attentive processing in visual search tasks can be directly linked to texture boundary processing as revealed by recent EEG studies on visual search.     

The radial bias: a different slant on visual orientation sensitivity in human and nonhuman primates

It is generally assumed that sensitivity to different stimulus orientations is mapped in a globally equivalent fashion across primate visual cortex, at a spatial scale larger than that of orientation columns. However, some evidence predicts instead that radial orientations should produce higher activity than other orientations, throughout visual cortex. Here, this radial orientation bias was robustly confirmed using (1) human psychophysics, plus fMRI in (2) humans and (3) behaving monkeys. In visual cortex, fMRI activity was at least 20% higher in the retinotopic representations of polar angle which corresponded to the radial stimulus orientations (relative to tangential). In a global demonstration of this, we activated complementary retinotopic quadrants of visual cortex by simply changing stimulus orientation, without changing stimulus location in the visual field. This evidence reveals a neural link between orientation sensitivity and the cortical retinotopy, which have previously been considered independent.     

Performance comparison of a neural network with human observers on a visual target detection task

An experiment is described which compares the performance of a neural network to human performance on a visual task which consists of detecting a target in a background image of correlated noise. A three-layer, feed-forward, multi-layer perceptron is trained to indicate the presence or absence of a target in images also presented to human observers. The basis for the comparison between the network and the human observers is the receiver operating characteristic (ROC) curve. Network performance is comparable to human performance for this particular task.     

Major claws make male fiddler crabs more conspicuous to visual predators: a test using human observers

One of the possible costs of the male fiddler crabs enlarged claw can be conspicuousness to predators. This hypothesis was tested using human observers as a model of visual predators. In the European fiddler crab, Uca tangeri Eydoux, the males' major claw is white contrasting with the orange-brownish colour of the carapace and of the feeding claw, and the mudflat background. The following morphotypes were created from close-up photographs taken in nature using an image processing software: male, male without claw, female, female with enlarged claw, male with enlarged claw of the same colour of the feeding claw, male with 75% sized claw, male with 50% sized claw. These morphotypes were then presented in a randomised order to students, using a psychology test software, which allows the measurement of response time in msec. The subjects were allowed to look at the images for an unlimited amount of time, until they detected the individual or until they decided to pass on to another image. Backgrounds (i.e. mudflat picture) without individuals were also presented as a control. Male crabs were detected significantly sooner than females. When we compared males with the claw removed with females with an enlarged claw added, the pattern is reversed and the latter are detected significantly faster. Thus, the enlarged claw seems to be the key feature that makes the individuals more conspicuous. Size and colour seem to be the main aspects of the claw's conspicuousness. The data of these experiments support the initial prediction of males being more conspicuous than females because of their enlarged claw. The possible costs and benefits of this trait, related to predation, are discussed.     

An overview of model observers

Nowadays, model observers have been used more and more for the objective quality assessment of medical images. Model observers have been developed from signal known exactly (SKE) task to signal known statistically (SKS) task, from single-slice (2D) to multi-slice (3D), in order to be more clinical relevant. In this paper, we give an overview of existing model observers up to date.     

A quantitative model of the functional architecture of human striate cortex with application to visual illusion and cortical texture analysis

Anatomical and physiological data from lower primates, and psychophysical data from humans, is used to construct a quantitative model of the local and global map structure (functional architecture) of human striate cortex. A series of successful estimates deriving from this model are reviewed, including a prediction for the width of human ocular dominance columns, which has recently been verified. A variety of perceptual phenomena are then discussed, from the point of view of cortical, rather than retinal, topography. It is suggested that the striate cortex may be viewed as a cyclopean retina whose non-linear map structure, summarized in terms of a concatenated complex logarithmic pattern, suggests insights into the nature of the Mackay complimentary image, the Frazer spiral, fortification illusions, and the relationship of the second order statistics of a visual stimulus to pre-attentive (textural) segmentation. Finally, the nature of neuronal representation is considered in the context of recent models of perceptual and cognitive function. It is suggested that anatomical re-mapping at successive stages of the CNS may provide a conceptual alternative to conventional single cell and connectionist models, and offers a viable approach towards a field theory of vision.     

Three processing characteristics of visual texture segmentation

Three mechanisms are outlined which are sufficient to determine texture segmentation or discrimination. They are: (1) convolution of detector profiles with the input image; (2) impletion, where the perceptual 'filling in' of the input surface occurs via a nonlinear filtering operation on each detector's output (3) grouping, where areas are segregated according to their differences in detector responses after impletion occurs. These mechanisms are compared with those proposed to occur in human visual texture discrimination.     

On perceptual analyzers underlying visual texture discrimination: Part I

We have found a class of feature detectors, based on the quasi-collinearity of dots, which result in visual texture discrimination even when second order statistics are equal. This degenerate counterexample to the Julesz conjecture on effortless texture discrimination has supplied the key to a simple theory of texture discrimination. Accordingly, effortless texture discrimination is based on two classses of perceptual detectors: Class A, those that measure differences in second-order (dipole) statistics; Class B, those that can still detect statistical differences in some features when second-order statistics are kept identical; for instance, the quasi-collinearity of adjacent dipoles. The difference thresholds (tuning curves) for the perceptual dipole and quasi-collinearity detectors have been determined. These texture pairs were generated by a method that creates micropatterns with iso-dipole duals from 4 disks. The extension of this 4-disk method to 5 and more disks with iso-dipole duals permits the search for other kinds of perceptual detectors and will be discussed in Part II.     

On perceptual analyzers underlying visual texture discrimination: Part II

In Part I Caelli and Julesz generated texture pairs of 4-disk micropatterns with identical dipole statistics. They found that this iso-dipole constraint could not prevent the quasi-collinearity of certain disk elements which, in turn, yielded effortless discrimination. They proposed two classes of perceptual analyzers to explain discrimination with these micropatern textures: Class A, corresponding to those which detect dipole differences; while Class B detectors, such as the quasi-collinear detector (QCD), acted when isodipole textures were presented. In this paper we show several new methods for generating iso-dipole textures with micropatterns consisting of 5 or more disks or non-disk shaped elements, and we report the discovery of two other Class B detectors, a corner detector (using a 6-disk method), and a closure detector (with 8–11 disk micropatterns). These QCD, corner, and closure detectors were verified by examining several hundred iso-dipole texture pairs. It appears that iso-dipole constraints make ineffective all other feature analyzers involved in effortless texture discrimination than the Class B types. These figural properties of collinearity, corners, and closure can be perceived without scrutiny and are precursors of form perception.     

Propagating waves in visual cortex: a large-scale model of turtle visual cortex

This article describes a large-scale model of turtle visual cortex that simulates the propagating waves of activity seen in real turtle cortex. The cortex model contains 744 multicompartment models of pyramidal cells, stellate cells, and horizontal cells. Input is provided by an array of 201 geniculate neurons modeled as single compartments with spike-generating mechanisms and axons modeled as delay lines. Diffuse retinal flashes or presentation of spots of light to the retina are simulated by activating groups of geniculate neurons. The model is limited in that it does not have a retina to provide realistic input to the geniculate, and the cortex and does not incorporate all of the biophysical details of real cortical neurons. However, the model does reproduce the fundamental features of planar propagating waves. Activation of geniculate neurons produces a wave of activity that originates at the rostrolateral pole of the cortex at the point where a high density of geniculate afferents enter the cortex. Waves propagate across the cortex with velocities of 4 m/ms to 70 m/ms and occasionally reflect from the caudolateral border of the cortex.     

Image quality in CT: From physical measurements to model observers

Evaluation of image quality (IQ) in Computed Tomography (CT) is important to ensure that diagnostic questions are correctly answered, whilst keeping radiation dose to the patient as low as is reasonably possible. The assessment of individual aspects of IQ is already a key component of routine quality control of medical x-ray devices. These values together with standard dose indicators can be used to give rise to ‘figures of merit’ (FOM) to characterise the dose efficiency of the CT scanners operating in certain modes. The demand for clinically relevant IQ characterisation has naturally increased with the development of CT technology (detectors efficiency, image reconstruction and processing), resulting in the adaptation and evolution of assessment methods. The purpose of this review is to present the spectrum of various methods that have been used to characterise image quality in CT: from objective measurements of physical parameters to clinically task-based approaches (i.e. model observer (MO) approach) including pure human observer approach. When combined together with a dose indicator, a generalised dose efficiency index can be explored in a framework of system and patient dose optimisation. We will focus on the IQ methodologies that are required for dealing with standard reconstruction, but also for iterative reconstruction algorithms. With this concept the previously used FOM will be presented with a proposal to update them in order to make them relevant and up to date with technological progress. The MO that objectively assesses IQ for clinically relevant tasks represents the most promising method in terms of radiologist sensitivity performance and therefore of most relevance in the clinical environment.     

Time estimation of fear cues in human observers

Previous research suggests that time judgments are a function of the affective properties of to-be-timed stimuli and that time judgments are longer for stimuli that are fear-inducing (e.g., [Hare, 1963] and [Watts and Sharrock, 1984]). The goals of the present study were twofold: to replicate the effect of a fear cue on time estimation, and to evaluate the mechanism underlying the effect. Seven stimulus durations in two different duration ranges (short: 250-1000 ms; long: 400-1600 ms) were employed in the bisection procedure. Adult human participants were exposed to two successive sessions, one each with the short and long range. Images from the International Affective Picture System (IAPS; Lang et al., 2008) that were rated on three scales including arousal and fear were presented as temporal stimuli. Three images that were rated high on fear and three rated low served as fear cues and neutral control images, respectively. Results indicated that for both ranges, judgments were longer for fear cues than for neutral images, and that the magnitude of the effect did not differ between ranges as measured by the bisection point. Application of scalar expectancy theory (SET; [Gibbon, 1977] and [Church, 1984]) to these results suggests that the fear effects were mediated by switch latency of an internal clock, rather than by clock speed.     

Visual perception of texture regularity: Conjoint measurements and a wavelet response-distribution model

Texture regularity, such as the repeating pattern in a carpet, brickwork or tree bark, is a ubiquitous feature of the visual world. The perception of regularity has generally been studied using multi-element textures in which the degree of regularity has been manipulated by adding random jitter to the elements’ positions. Here we used three-factor Maximum Likelihood Conjoint Measurement (MLCM) for the first time to investigate the encoding of regularity information under more complex conditions in which element spacing and size, in addition to positional jitter, were manipulated. Human observers were presented with large numbers of pairs of multi-element stimuli with varying levels of the three factors, and indicated on each trial which stimulus appeared more regular. All three factors contributed to regularity perception. Jitter, as expected, strongly affected regularity perception. This effect of jitter on regularity perception is strongest at small element spacing and large texture element size, suggesting that the visual system utilizes the edge-to-edge distance between elements as the basis for regularity judgments. We then examined how the responses of a bank of Gabor wavelet spatial filters might account for our results. Our analysis indicates that the peakedness of the spatial frequency (SF) distribution, a previously favored proposal, is insufficient for regularity encoding since it varied more with element spacing and size than with jitter. Instead, our results support the idea that the visual system may extract texture regularity information from the moments of the SF-distribution across orientation. In our best-performing model, the variance of SF-distribution skew across orientations can explain 70% of the variance of estimated texture regularity from our data, suggesting that it could provide a candidate read-out for perceived regularity.     

A space-variant filter model of texture segregation: Parameter adjustment guided by psychophysical data

 This article presents a space-variant version of a standard spatial filter model of texture segregation of the “back-pocket” type (i.e., two filter layers with an intermediate pointwise nonlinearity). The model was tested with psychophysical data from experiments with line textures in which target lines differed in orientation from background lines. The textures were presented briefly and then masked. Segregation performance was evaluated along the horizontal meridian up to retinal eccentricities of about 10 deg. Data are reported from two experiments with different line densities (Kehrer 1989) and two experiments with different orientation contrasts between target lines and background lines (Kehrer 1990). Segregation performance proved to depend strongly on these texture variations, and it peaked several degrees from fixation in all cases. The filter model provided satisfactory predictions of experimental data when model parameters were adjusted appropriately. It is concluded (1) that filter models defined in strictly spatial terms (i.e., without temporal properties) offer a sufficient framework to account for the psychophysical data and (2) that the particular course of the performance curve (i.e., the performance peak outside the central region) must be attributed to the characteristics of second-layer filters. Received: 28 June 2001 / Accepted in revised form: 10 October 2002 / Published online: 13 February 2003 Correspondence to: L. Kehrer (e-mail: lothar.kehrer@uni-bielefeld.de) Acknowledgements. This work was supported by Grant Ke 388/3-2 from the Deutsche Forschungsgemeinschaft (DFG). We wish to thank Jonathan Harrow for improving the English text and an anonymous reviewer for many constructive comments.     

Directional inhibition: a new slant on an old question

In this issue of Neuron, Priebe and Ferster describe the direction selectivity and spatiotemporal organization of excitatory and inhibitory inputs to direction-selective simple cells in cat visual cortex. Their most surprising finding is that inhibition shows the same preferred direction as excitation.     

A computational model of the error detector of human visual accommodation

A mathematical model is proposed for the error detector of the human visual accommodative system. The model supposes that the accommodative error detector derives both the direction and the magnitude of the accommodative error from naturally-occuring oscillations of the lens and their effects on retinal-image contrast. Differential operators take the first derivatives of two time varying functions: lens power and retinal-image contrast. Directional information is obtained by comparing the signs of these two derivatives and magnitude information is obtained by comparing their amplitudes.Research conducted at the School of Optometry, University of California, BerkeleySupported by National Eye Institute grant EYO-3532-04(C.S.) and National Institutes of Health core grant # 1-445420-32011