Preattentive Texture Discrimination Research Articles

Texture‐based image segmentation using neutrosophic clustering

This study presents an effective segmentation method which is based on neutrosophic clustering with the integration of texture features for images. The proposed method transforms the image into the neutrosophic domain and then extracts the texture features using analogies of human preattentive texture discrimination mechanisms. Finally, the neutrosophic clustering is employed to segment the images. This method can handle the indeterminacy of pixels to have strong clusters and to perform segmentation effectively with the noisy images. Experiments are performed with various types of natural and medical images to exhibit the performance of proposed segmentation method. The evaluation of proposed method has been done with other segmentation methods to measure its performance which shows its robustness for noisy and textured images.

Different Processing Phases for Features, Figures, and Selective Attention in the Primary Visual Cortex

Our visual system imposes structure onto images that usually contain a diversity of surfaces, contours, and colors. Psychological theories propose that there are multiple steps in this process that occur in hierarchically organized regions of the cortex: early visual areas register basic features, higher areas bind them into objects, and yet higher areas select the objects that are relevant for behavior. Here we test these theories by recording from the primary visual cortex (area V1) of monkeys. We demonstrate that the V1 neurons first register the features (at a latency of 48 ms), then segregate figures from the background (after 57 ms), and finally select relevant figures over irrelevant ones (after 137 ms). We conclude that the psychological processing stages map onto distinct time episodes that unfold in the visual cortex after the presentation of a new stimulus, so that area V1 may contribute to all these processing steps.

Intrinsic two-dimensional features as textons.

We suggest that intrinsic two-dimensional (i2D) features, computationally defined as the outputs of nonlinear operators that model the activity of end-stopped neurons, play a role in preattentive texture discrimination. We first show that for discriminable textures with identical power spectra the predictions of traditional models depend on the type of nonlinearity and fail for energy measures. We then argue that the concept of intrinsic dimensionality, and the existence of end-stopped neurons, can help us to understand the role of the nonlinearities. Furthermore, we show examples in which models without strong i2D selectivity fail to predict the correct ranking order of perceptual segregation. Our arguments regarding the importance of i2D features resemble the arguments of Julesz and co-workers regarding textons such as terminators and crossings. However, we provide a computational framework that identifies textons with the outputs of nonlinear operators that are selective to i2D features.

Magnetic and electrical responses of the human brain to texture-defined form and to textons.

1. We searched for a neurophysical correlate of preattentive texture discrimination by recording magnetic and electric evoked responses from the human brain during the first few hundred milliseconds following the presentation of texture-defined (TD) checkerboard form. The only two textons that changed when the TD checkerboard appeared or disappeared were the local orientation and line termination textons. (Textons are conspicuous local features within a texture pattern). 2. Our evidence that the magnetic response to TD form cannot be explained in terms of responses to the two associated textons is as follows: 1) by dissociating the two responses we showed that the magnetic response to TD form is almost entirely independent of the magnetic response to the local orientation texton; 2) a further distinction between the two responses is that their distributions over the head are different; and 3) the magnetic response to TD form differs from the magnetic response to the line termination texton in both distribution over the head and waveform. We conclude that this evidence identifies the existence of a brain response correlate of preattentive texture discrimination. 3. We also recorded brain responses to luminance-defined (LD) checkerboard form. Our grounds for concluding that magnetic brain responses to the onset of checkerboard form are generated by different and independent neural systems for TD and LD form are as follows: 1) magnetic responses to the onset of TD form and LD form had different distributions over the skull, had different waveforms, and depended differently on check size; and 2) the waveform of the response to superimposed TD and LD checks closely approximated the linear sum of responses to TD checks and LD checks alone. 4. One possible explanation for the observed differences between the magnetic and electric evoked responses is that responses to both onset and offset of TD form predominantly involve neurons aligned parallel to the skull, whereas that is not the case for responses to LD form.

Early vision and focal attention

At the thirty-year anniversary of the introduction of the technique of computer-generated random-dot stereograms and random-dot cinematograms into psychology, the impact of the technique on brain research and on the study of artificial intelligence is reviewed. The main finding---that stereoscopic depth perception (stereopsis), motion perception, and preattentive texture discrimination are basically bottom-up processes, which occur without the help of the top-down processes of cognition and semantic memory---greatly simplifies the study of these processes of early vision and permits the linking of human perception with monkey neurophysiology. Particularly interesting are the unexpected findings that stereopsis (assumed to be local) is a global process, while texture discrimination (assumed to be a global process, governed by statistics) is local, based on some conspicuous local features (textons). It is shown that the top-down process of "shape (depth) from shading" does not affect stereopsis, and some of the models of machine vision are evaluated. The asymmetry effect of human texture discrimination is discussed, together with recent nonlinear spatial filter models and a novel extension of the texton theory that can cope with the asymmetry problem. This didactic review attempts to introduce the physicist to the field of psychobiology and its problems---including metascientific problems of brain research, problems of scientific creativity, the state of artificial intelligence research (including connectionist neural networks) aimed at modeling brain activity, and the fundamental role of focal attention in mental events.

Computing with Arrays of Coupled Oscillators: An Application to Preattentive Texture Discrimination

Recent experimental findings (Gray et al. 1989; Eckhorn et al. 1988) seem to indicate that rapid oscillations and phase-lockings of different populations of cortical neurons play an important role in neural computations. In particular, global stimulus properties could be reflected in the correlated firing of spatially distant cells. Here we describe how simple coupled oscillator networks can be used to model the data and to investigate whether useful tasks can be performed by oscillator architectures. A specific demonstration is given for the problem of preattentive texture discrimination. Texture images are convolved with different sets of Gabor filters feeding into several corresponding arrays of coupled oscillators. After a brief transient, the dynamic evolution in the arrays leads to a separation of the textures by a phase labeling mechanism. The importance of noise and of long range connections is briefly discussed.

Preattentive texture discrimination with early vision mechanisms

We present a model of human preattentive texture perception. This model consists of three stages: (1) convolution of the image with a bank of even-symmetric linear filters followed by half-wave rectification to give a set of responses modeling outputs of V1 simple cells, (2) inhibition, localized in space, within and among the neural-response profiles that results in the suppression of weak responses when there are strong responses at the same or nearby locations, and (3) texture-boundary detection by using wide odd-symmetric mechanisms. Our model can predict the salience of texture boundaries in any arbitrary gray-scale image. A computer implementation of this model has been tested on many of the classic stimuli from psychophysical literature. Quantitative predictions of the degree of discriminability of different texture pairs match well with experimental measurements of discriminability in human observers.

Texture segregation with luminance and chromatic contrast

Preattentive texture discrimination was investigated using low spatial frequency texture elements. The contrast between the texture elements and the background was either purely luminance or purely chromatic, or some combination of both these types of contrast. The threshold to discriminate correctly the location of a different textured region was obtained from each subject, as was each subject's threshold to detect the elements of the texture. Using the ratio of texture to element detection as a measure of the effectiveness of texture discrimination, little difference could be found between the perception of luminance or chromatic texture. However, there were large and significant variations among subjects with otherwise normal colour vision.

Models for preattentive texture discrimination: Fourier analysis and local feature processing in a unified framework

Spatial frequency analysis and local feature analysis may be considered to be examples of a class of models for texture discrimination. In this theoretical framework, texture discrimination relies on differences in the distribution of responses generated in linear receptive fields placed randomly on the texture. If the set of receptive fields is taken to be a collection of gratings, spatial-frequency analysis is recovered. If the set of receptive-fields is taken to be a collection of local feature templates, a corresponding local-feature model is recovered. In order to test such models, it is necessary to construct distinct texture pairs that elicit similar distributions of responses for all of the postulated receptive field profiles: the model prediction is that such textures are not discriminable. A method is provided for construction of such textures which test generic models within this framework. This framework includes not only strict Fourier analysis, but also models which postulate a collection of arbitrarily-shaped local feature detectors, and models which postulate both Fourier analysis and local feature detection.

Texture discrimination at different eccentricities.

Differences in preattentive texture discrimination between central vision and peripheral vision were studied with textures composed of random dots. The subject had to discriminate between two textures whose first-order statistics were kept identical but whose second-order statistics were different. For textures of constant retinal size the discrimination was easy in central vision, but the decrease of visual acuity with increasing eccentricity made the textures unresolvable in peripheral vision. When the textures were scaled by the cortical magnification factor derived from the frequency of retinal ganglion cells so that the calculated neural representations of the textures became similar at different eccentricities, texture discrimination became independent of visual field location. This indicates that preattentive texture discrimination based on differences in second-order statistics of random dots operates similarly in central vision and peripheral vision.

Local structure analyzers as determinants of preattentive pattern discrimination.

Contemporary literature suggests that preattentive texture or pattern discrimination is induced by differences between local structure features or "textons." This paper presents a model for the description of such local structure features based on the computation of local autocorrelations within the image. By means of this structure model a measure of structure dissimilarity is introduced. Experiments have been carried out to test a hypothesized relation between the detectability of a target pattern in a field of background patterns and the value of the structure dissimilarity measure. The experimental results show that it seems justified to relate, in a quantitative way, the detectability of the target pattern to the value of the structure dissimilarity measure.

Texture discrimination by Gabor functions

A 2D Gabor filter can be realized as a sinusoidal plane wave of some frequency and orientation within a two dimensional Gaussian envelope. Its spatial extent, frequency and orientation preferences as well as bandwidths are easily controlled by the parameters used in generating the filters. However, there is an "uncertainty relation" associated with linear filters which limits the resolution simultaneously attainable in space and frequency. Daugman (1985) has determined that 2D Gabor filters are members of a class of functions achieving optimal joint resolution in the 2D space and 2D frequency domains. They have also been found to be a good model for two dimensional receptive fields of simple cells in the striate cortex (Jones 1985; Jones et al. 1985). The characteristic of optimal joint resolution in both space and frequency suggests that these filters are appropriate operators for tasks requiring simultaneous measurement in these domains. Texture discrimination is such a task. Computer application of a set of Gabor filters to a variety of textures found to be preattentively discriminable produces results in which differently textured regions are distinguished by first-order differences in the values measured by the filters. This ability to reduce the statistical complexity distinguishing differently textured region as well as the sensitivity of these filters to certain types of local features suggest that Gabor functions can act as detectors of certain "texton" types. The performance of the computer models suggests that cortical neurons with Gabor like receptive fields may be involved in preattentive texture discrimination.

Contrast and conspicuity in preattentive texture perception.

Preattentive texture discrimination is thought to occur as the result of differences in the first-order statistics of conspicuous local features known as textons (Julesz, 1984a, 1984b). The present study investigated discrimination as a function of two texton properties: (1) blob contrast and (2) texton conspicuity as a function of contrast polarity. The results demonstrated that preattentive texture discrimination is a function of differences in contrast distribution, and that conspicuity varies with texton type.

Texton gradients: the texton theory revisited.

A brief outline of the texton theory was given in several review papers (Julesz and Bergen 1983; Julesz 1984a, b, 1985) without going into details. Here a more complete version of the texton theory is presented, with emphasis on the critical distances within which the density of textons is determined by the preattentive system. Particularly some recent findings by Sagi and Julesz (1985a, b) influenced the current version of the texton theory. The stimuli are restricted to drawings composed of line segments that permit a precise definition of neighborhoods and distances to which the line segments and texture elements have to be confined in order to quality for preattentive texture discrimination. These critical distances and the aperture of focal attention are scaled by the average size of the texture elements. Furthermore, it is stressed that even when the stimuli are restricted to line segments, the blobs outlined by line segments behave like textons. The preattentive system ignores the exact shape of these blobs, but is sensitive to their average width, length, and orientation.

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 that globally 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 on local conspicuous features called textons (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.

Textons, the elements of texture perception, and their interactions.

Research with texture pairs having identical second-order statistics has revealed that the pre-attentive texture discrimination system cannot globally process third- and higher-order statistics, and that discrimination is the result of a few local conspicuous features, called textons. It seems that only the first-order statistics of these textons have perceptual significance, and the relative phase between textons cannot be perceived without detailed scrutiny by focal attention.