A theory of preattentive texture discrimination based on first-order statistics of textons

The many indistinguishable texture pairs having identical second-, but different third- and higher-order statistics, led to the conjecture thatglobally the preattentive texture discrimination system cannot process statistical parameters of third- or higher-order. Thus in cases when iso-second-order textures yield discrimination this must be based onlocal conspicuous features calledtextons (Julesz, 1980). Here it is shown that globally even second-order statistical parameters, such as autocorrelation, cannot be processed by the textural system, and texture discrimination is solely the result of first-order statistics (density) of textons. It is also shown that the perceivable distance of statistical constraints (coherence distance) in densely packed stochastic textures is very short, four dots or less. As of now, only three texton classes were found: color, elongated blobs (line segments) of given width, orientation, and length, and the terminators (end-points) of these elongated blobs. The strength of these textons is demonstrated by several examples.     

Spatial nonlinearities in the instantaneous perception of textures with identical power spectra

In 1962, Julesz observed that texture pairs with identical second-order statistics but different third- and higher-order statistics were usually not discriminable without scrutiny. Since second-order (dipole) statistics determine the autocorrelation functions and hence the power spectra, this observation also meant that in preattentive perception of texture the phase (position) spectra were ignored. In the last two decades many new classes of texture pairs with identical power spectra have been invented that were not effortlessly discriminable; however, recently (Caelli & Julesz 1978; Caelli et al. 1978; Julesz et al. 1978) several counterexamples were found. In these texture pairs with identical power spectra some local structures of 'quasi-collinearity', 'corner', 'closure' and 'granularity' yielded strong discrimination. These features can be regarded as the fundamental building blocks of form, that is, the essential nonlinearities of the preattentive perceptual system. Here, it will be shown that these counter-examples are not independent of each other, but can be described by two elementary units; bars (line segments) and their terminators. Furthermore, the preattentive texture perception system can count the number of terminators but ignores their positions.     

A model for the perception of curves in dot figures: The role of local salience of “virtual lines”

In many models of visual information processing the notion of a virtual line or dipole is introduced in order to represent the configurational information, notably length and relative orientation, between identical figure elements in figures with discrete elements. Virtual lines have proven to be very useful in predicting perceptual phenomena (Julesz et al. 1973; Stevens 1978). In the present study, virtual lines are utilized in a model which aims to predict the perception of (dotted) curves in dot figures. Clearly many possible curves, formed by adjacent virtual lines, can be constructed within a set of dots. It is proposed that already at the local level of the virtual lines each line has a perceptual salience which results from the function induced by the global dot figure. It is this local line salience or connectivity that directs further processing and determines the curves to be seen in a dot figure. The model presented is an information processing model with a clear modular design. It entails three successive levels of representation. First image functions are derived through a convolution of the input with gaussian distribution functions. Next, a discrete internal representation is extracted from the image function consisting of two primitives; blobs, representing the dots, and virtual lines, representing pairwise relations between blobs. The attributes of the blobs are their positions in the image plane, while those of the virtual lines are length, relative orientation and connectivity. At the third level, the discrete internal representation is used to predict the perceived curves. It is shown that the model has advantages over other approaches, e.g. autocorrelation and network models.     

The central performance drop in texture segmentation: a simulation based on a spatial filter model

 This article presents a computational model of early visual information processing that attempts to account for the central performance drop (CPD) in texture segmentation. CPD is the finding that detection performance on short stimulus displays of line textures using orientation differences to set off the target is not maximal at the foveal center but in parafoveal areas. A comparison between a simulation and psychophysical experimental data supported the assumption that the CPD may be explained by properties of spatial frequency channels whose band-pass filter characteristics are not constant over the retina but differ with eccentricity in a defined manner. The model provided satisfactory predictions of experimental data based on densely or widely spaced line elements in texture fields. It is concluded that preattentive texture analysis might be performed by a relatively small number of simple spatial filters. Received: 14 November 1996 / Accepted in revised form: 3 June 1997     

Gabor filters as texture discriminator

The present paper presents a model for texture discrimination based on Gabor functions. In this model the Gabor power spectrum of the micropatterns corresponding to different textures is calculated. A function that measures the difference between the spectrum of two micropatterns is introduced and its values are correlated with human performance in preattentive detection tasks. In addition, a two stage algorithm for texture segregation is presented. In the first stage the input image is transformed via Gabor filters into a representation image that allows discrimination between features by means of intensity differences. In the second stage the borders between areas of different textures are found using a Laplacian of Gaussian operator. This algorithm is sensitive to energy differences, rotation and spatial frequency and is insensitive to local translation. The model was tested by means of several simulations and was found to be in good correlation with known psychophysical characteristics as texton based texture segregation and micropattern density sensitivity. However, this simple model fails to predict human performance in discrimination tasks based on differences in the density of terminators. In this case human performance is better than expected.     

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.     

Line segregation

Six experiments investigated the preattentive segregation of line-like patterns composed of discrete elements in a background of distractors. The results indicate that other factors in addition to spatial density influence line segregation. Edge alignment, edge length and principal axis orientation also affect line segregation. Differences in the outputs of Gabor filters fail to account for the perceived segregation of the lines. Possible models of line segregation based on element grouping, feature density and search are briefly discussed.     

Prey-capture in the African clawed toad (Xenopus laevis): comparison of turning to visual and lateral line stimuli

Separately delivered visual and lateral line stimuli elicit similar but not identical orientation and approach by intact, sighted Xenopus. Response frequencies for visual stimuli declined sharply for distant or caudal stimuli while those for lateral line stimuli changed little. Turn angles correlated highly with stimulus angles but were smaller on average, so regression slopes were less than one. Regression slopes were smaller for visual than for lateral line stimuli, but this apparent difference was due to different distributions of stimulus distance interacting with the toad’s rotation center. Errors in final headings, most often under-rotations, did not differ by modality. Frequencies of lunges and arm capture movements were higher for visual stimuli both overall and especially for rostral proximal stimuli. The results demonstrate accurate orientation by sighted Xenopus to visual and lateral line stimuli; they are consistent with expectations based on in-register tectal maps. Orientation to lateral line stimuli is similar to previous results with blinded animals, revealing no heightened acuity in the latter. Modality differences indicate that the lateral line system is better for omnidirectional orientation and approach to distant stimuli whereas the visual system is more attuned to nearby rostral stimuli and more apt to mediate strikes.Electronic Supplementary Material Supplementary material is available to authorised users in the online version of this article at .     

Short-range limitation on detection of feature differences

We studied the ability of observers to detect the presence of a clearly visible line segment against a background of line segments of different orientation. As we increase the number (density) of these background lines, we find that detectability does not behave monotonically. Adding a small number of background lines decreases detectability but if adjacent line segments are permitted to fall in close range, a further increase of background lines improves performance which eventually reaches a constant level. This suggests that detection of feature differences involves a short-range process. The range of this process is about two degrees or twice the length of the line segments used. Thus texture-gradients between different elements are only formed if the distance between these elements is not much larger than the average element size.     

Backward masking is not required to elicit the central performance drop

In some circumstances, texture discrimination performance peaks in the parafovea rather than at the fovea. Kehrer (1987) referred to this phenomenon as the central performance drop (CPD). In most studies showing the CPD, task performance has been limited by a backward mask. Morikawa (2000) has argued that in these studies the backward mask was critical to the emergence of the CPD. In three studies we use textures comprising left and right oblique line segments and limit performance by manipulating the orientation variability within the foreground and background textures. Using this method we demonstrate that significant CPDs emerge whether or not there is a backward mask. We conclude that in past studies of the CPD the backward mask functioned primarily as a source of spatial noise and that its temporal relation to the texture display is not critical to the emergence of the CPD.     

Selective and divided attention in animals

This article reviews some of the research on attentional processes in animals. In the traditional approach to selective attention, it is proposed that in addition to specific response attachments, animals also learn something about the dimension along which the stimuli fall (e.g., hue, brightness, or line orientation). More recently, there has been an attempt to find animal analogs to methodologies originally applied to research with humans. One line of research has been directed to the question of whether animals can locate a target among distracters faster if they are prepared for the presentation of the target (search image and priming). In the study of search image, the target is typically a food item and the cue consists of previous trials on which the same target is presented. In research on priming effects, the cue is typically different from the target but is a good predictor of its occurrence. The study of preattentive processes shows that perceptually, certain stimuli stand out from distracters better than others, depending not only on characteristics of the target relative to the distracters, but also on relations among the distracters. Research on divided attention is examined with the goal of determining whether an animal can process two elements of a compound sample with the same efficiency as one. Taken together, the reviewed research indicates that animals are capable of centrally (not just peripherally) attending to selective aspects of a stimulus display.     

Isotropic integration of binocular disparity and relative motion in the perception of three-dimensional shape

Richards (1985) showed that veridical three-dimensional shape may be recovered from the integration of binocular disparity and retinal motion information, but proposed that this integration may only occur for horizontal retinal motion. Psychophysical evidence supporting the combination of stereo and motion information is limited to the case of horizontal motion (Johnston et al., 1994), and has been criticised on the grounds of potential object boundary cues to shape present in the stimuli. We investigated whether veridical shape can be recovered under more general conditions. Observers viewed cylinders that were defined by binocular disparity, two-frame motion or a combination of disparity and motion, presented at simulated distances of 30 cm, 90 cm or 150 cm. Horizontally and vertically oriented cylinders were rotated about vertical and horizontal axes. When rotation was about the cylinder's own axis, no boundary cues to shape were introduced. Settings were biased for the disparity and two-frame motion stimuli, while more veridical shape judgements were made under all conditions for combined cue stimuli. These results demonstrate that the improved perception of three-dimensional shape in these stimuli is not a consequence of the presence of object boundary cues, and that the combination of disparity and motion is not restricted to horizontal image motion.     

Preattentive extraction of abstract auditory rules in speech sound stream: a mismatch negativity study using lexical tones

Background

Extraction of linguistically relevant auditory features is critical for speech comprehension in complex auditory environments, in which the relationships between acoustic stimuli are often abstract and constant while the stimuli per se are varying. These relationships are referred to as the abstract auditory rule in speech and have been investigated for their underlying neural mechanisms at an attentive stage. However, the issue of whether or not there is a sensory intelligence that enables one to automatically encode abstract auditory rules in speech at a preattentive stage has not yet been thoroughly addressed.

Methodology/Principal Findings

We chose Chinese lexical tones for the current study because they help to define word meaning and hence facilitate the fabrication of an abstract auditory rule in a speech sound stream. We continuously presented native Chinese speakers with Chinese vowels differing in formant, intensity, and level of pitch to construct a complex and varying auditory stream. In this stream, most of the sounds shared flat lexical tones to form an embedded abstract auditory rule. Occasionally the rule was randomly violated by those with a rising or falling lexical tone. The results showed that the violation of the abstract auditory rule of lexical tones evoked a robust preattentive auditory response, as revealed by whole-head electrical recordings of the mismatch negativity (MMN), though none of the subjects acquired explicit knowledge of the rule or became aware of the violation.

Conclusions/Significance

Our results demonstrate that there is an auditory sensory intelligence in the perception of Chinese lexical tones. The existence of this intelligence suggests that the humans can automatically extract abstract auditory rules in speech at a preattentive stage to ensure speech communication in complex and noisy auditory environments without drawing on conscious resources.     

Topography of genetic loci in tissue samples: towards new diagnostic tool using interphase FISH and high-resolution image analysis techniques.

Using single and dual colour fluorescence in situ hybridisation (FISH) combined with image analysis techniques the topographic characteristics of genes and centromeres in nuclei of human colon tissue cells were investigated. The distributions of distances from the centre-of-nucleus to genes (centromeres) and from genes to genes (centromeres to centromeres) were studied in normal colon tissue cells found in the neighbourhood of tumour samples, in tumour cell line HT-29 and in promyelocytic HL-60 cell line for comparison. Our results show that the topography of genetic loci determined in 3D-fixed cell tissue corresponds to that obtained for 2D-fixed cells separated from the tissue. The distributions of the centre-of-nucleus to gene (centromere) distances and gene to gene (centromere to centromere) distances and their average values are different for various genetic loci but similar for normal colon tissue cells, HT-29 colon tumour cell line and HL-60 promyelocytic cell line. It suggests that the arrangement of genetic loci in cell nucleus is conserved in different types of human cells. The investigations of trisomic loci in HT-29 cells revealed that the location of the third genetic element is not different from the location of two homologues in diploid cells. We have shown that the topographic parameters used in our experiments for different genetic elements are not tissue or tumour specific. In order to validate high-resolution cytometry for oncology, further investigations should include more precise parameters reflecting the state of chromatin in the neighbourhood of critical oncogenes or tumour suppresser genes.     

Perceptual segregation and retinal position

This study is concerned with the dependence of perceptual segregation performance on the retinal position at which the performance is evaluated. Segregation performance consisted in detecting a target texture composed of line elements of constant length and orientation embedded in a background texture. The background texture was made up of the same line elements as the target texture but the background line elements were set at 90 deg to the target elements. Results showed that, at least for 40- and 50-ms presentation times, this task could be much more effectively completed outside the fovea centralis than within this area. These findings indicate that extrafoveal areas of the retina may make a significant and previously underestimated contribution to perceptual segregation.     

Defining linear segments in protein structure

The analysis of protein structure using secondary structure line segments has been widely used in many structure analysis and prediction methods over the past 20 years. Its use in methods that compare protein structures at this level of representation is becoming more important as an increasing number of protein structures become determined through structural genomic programmes. The standard method used to define line segments is to fit an axis through each secondary structure element. This approach has difficulties, however, both with inconsistent definitions of secondary structure and the problem of fitting a single straight line to a bent structure. The procedure described here avoids these problems by finding a set of line segments independently of any external secondary structure definition. This allows the segments to be used as a novel basis for secondary structure definition by taking the average rise/residue along each axis to characterise the segment. This practice has the advantage that secondary structures are described by a single (continuous) value that is not restricted to the conventional classes of alpha-helix, 310 and beta-strand. This latter property allows structures without "classic" secondary structures to be encoded as line segments that can be used in comparison algorithms. When compared over a large number of pairs of homologous proteins, the current method was found to be slightly more consistent than a widely used method based on hydrogen bonds.