Perceived contrast in complex images

Suprathreshold contrast perception of resolvable high spatial frequencies remain intact in keratoconus

Contrast detection thresholds are elevated with optical quality loss in keratoconus. This study hypothesized that suprathreshold contrast perception is also impaired in keratoconus, with the impairment being predictable from the pattern of loss in threshold-level performance. Contrast detection thresholds were determined across a range of spatial frequencies in 12 cases with mild to severe keratoconus and 12 age-similar controls. These values were used to predict the contrast needed to achieve perceptual matches between reference and test spatial frequency pairs (peak of CSF Vs. 0.3x, 0.5x, 2x or 3x spatial frequency from the peak) for stimuli at 10% and 50% suprathreshold contrast. Contrast thresholds predicted a 1.5 to 6.7-fold increase in the test pattern's contrast to obtain a perceptual match with the reference pattern in keratoconus, relative to controls. Contrary to predictions, the empirical data of contrast matches between test and reference patterns were similar for higher than peak spatial frequencies at both contrast levels. However, as predicted, test patterns required higher contrast than the reference pattern for a perceptual match for lower than peak spatial frequencies. These results were similar to controls and invariant of disease severity, interocular asymmetry and short-term changes in optical quality. Unlike thresholds, suprathreshold contrast perception of resolvable high spatial frequencies appears immune to optical quality losses in keratoconus. These results are discussed in the context of the prevailing models of contrast constancy in healthy humans. Breakdown of contrast constancy at lower than peak spatial frequencies may reflect the properties of the testing paradigm employed here.

Natural Contrast Statistics Facilitate Human Face Categorization

The ability to detect faces in the environment is of utmost ecological importance for human social adaptation. While face categorization is efficient, fast and robust to sensory degradation, it is massively impaired when the facial stimulus does not match the natural contrast statistics of this visual category, i.e., the typically experienced ordered alternation of relatively darker and lighter regions of the face. To clarify this phenomenon, we characterized the contribution of natural contrast statistics to face categorization. Specifically, 31 human adults viewed various natural images of nonface categories at a rate of 12 Hz, with highly variable images of faces occurring every eight stimuli (1.5 Hz). As in previous studies, neural responses at 1.5 Hz as measured with high-density electroencephalography (EEG) provided an objective neural index of face categorization. Here, when face images were shown in their naturally experienced contrast statistics, the 1.5-Hz face categorization response emerged over occipito-temporal electrodes at very low contrast [5.1%, or 0.009 root-mean-square (RMS) contrast], quickly reaching optimal amplitude at 22.6% of contrast (i.e., RMS contrast of 0.041). Despite contrast negation preserving an image's spectral and geometrical properties, negative contrast images required twice as much contrast to trigger a face categorization response, and three times as much to reach optimum. These observations characterize how the internally stored natural contrast statistics of the face category facilitate visual processing for the sake of fast and efficient face categorization.

Contrast discrimination in images of natural scenes

Contrast discrimination determines the threshold contrast required to distinguish between two suprathreshold visual stimuli. It is typically measured using sine-wave gratings. We first present a modification to Barten's semi-mechanistic contrast discrimination model to account for spatial frequency effects and demonstrate how the model can successfully predict visual thresholds obtained from published classical contrast discrimination studies. Contrast discrimination functions are then measured from images of natural scenes, using a psychophysical paradigm based on that employed in our previous study of contrast detection sensitivity. The proposed discrimination model modification is shown to successfully predict discrimination thresholds for structurally very different types of natural image stimuli. A comparison of results shows that, for normal contrast levels in natural scene viewing, contextual contrast detection and discrimination are approximately the same and almost independent of spatial frequency within the range of 1-20 c/deg. At higher frequencies, both sensitivities decrease in magnitude due to optical limitations of the eye. The results are discussed in relation to current image quality models.

Conjoint measurement of perceived transparency and perceived contrast in variegated checkerboards

One fundamental question in vision research is how the retinal input is segmented into perceptually relevant variables. A striking example of this segmentation process is transparency perception, in which luminance information in one location contributes to two perceptual variables: the properties of the transparent medium itself and of what is being seen in the background. Previous work by Robilotto et al. (2002, 2004) suggested that perceived transparency is closely related to perceived contrast, but how these two relate to retinal luminance has not been established. Here we studied the relationship between perceived transparency, perceived contrast, and image luminance using maximum likelihood conjoint measurement (MLCM). Stimuli were rendered images of variegated checkerboards that were composed of multiple reflectances and partially covered by a transparent overlay. We systematically varied the transmittance and reflectance of the transparent medium and measured perceptual scales of perceived transparency. We also measured scales of perceived contrast using cut-outs of the transparency stimuli that did not contain any geometrical cues to transparency. Perceptual scales for perceived transparency and contrast followed a remarkably similar pattern across observers. We tested the empirically observed scales against predictions from various contrast metrics and found that perceived transparency and perceived contrast were equally well predicted by a metric based on the logarithm of Michelson or Whittle contrast. We conclude that judgments of perceived transparency and perceived contrast are likely to be supported by a common mechanism, which can be computationally captured as a logarithmic contrast.

Spatial structure, phase, and the contrast of natural images

The sensitivity of the human visual system is thought to be shaped by environmental statistics. A major endeavor in vision science, therefore, is to uncover the image statistics that predict perceptual and cognitive function. When searching for targets in natural images, for example, it has recently been proposed that target detection is inversely related to the spatial similarity of the target to its local background. We tested this hypothesis by measuring observers' sensitivity to targets that were blended with natural image backgrounds. Targets were designed to have a spatial structure that was either similar or dissimilar to the background. Contrary to masking from similarity, we found that observers were most sensitive to targets that were most similar to their backgrounds. We hypothesized that a coincidence of phase alignment between target and background results in a local contrast signal that facilitates detection when target-background similarity is high. We confirmed this prediction in a second experiment. Indeed, we show that, by solely manipulating the phase of a target relative to its background, the target can be rendered easily visible or undetectable. Our study thus reveals that, in addition to its structural similarity, the phase of the target relative to the background must be considered when predicting detection sensitivity in natural images.

What is visible across the visual field?

It is sometimes claimed that because the resolution and sensitivity of visual perception are better in the fovea than in the periphery, peripheral vision cannot support the same kinds of colour and sharpness percepts as foveal vision. The fact that a scene nevertheless seems colourful and sharp throughout the visual field then poses a puzzle. In this study, I use a detailed model of human spatial vision to estimate the visibility of certain properties of natural scenes, including aspects of colourfulness, sharpness, and blurriness, across the visual field. The model is constructed to reproduce basic aspects of human contrast and colour sensitivity over a range of retinal eccentricities. I apply the model to colourful, complex natural scene images, and estimate the degree to which colour and edge information are present in the model's representation of the scenes. I find that, aside from the intrinsic drift in the spatial scale of the representation, there are not large qualitative differences between foveal and peripheral representations of 'colourfulness' and 'sharpness'.

Reversing the Luminance Polarity of Control Faces: Why Are Some Negative Faces Harder to Recognize, but Easier to See?

Control stimuli are key for understanding the extent to which face processing relies on holistic processing, and affective evaluation versus the encoding of low-level image properties. Luminance polarity (LP) reversal combined with face inversion is a popular tool for severely disrupting the recognition of face controls. However, recent findings demonstrate visibility-recognition trade-offs for LP-reversed faces, where these face controls sometimes appear more salient despite being harder to recognize. The present report brings together findings from image analysis, simple stimuli, and behavioral data for facial recognition and visibility, in an attempt to disentangle instances where LP-reversed control faces are associated with a performance bias in terms of their perceived salience. These findings have important implications for studies of subjective face appearance, and highlight that future research must be aware of behavioral artifacts due to the possibility of trade-off effects.

Suppression durations for facial expressions under breaking continuous flash suppression: effects of faces' low-level image properties

Perceptual biases for fearful facial expressions are observed across many studies. According to the low-level, visual-based account of these biases, fear expressions are advantaged in some way due to their image properties, such as low spatial frequency content. However, there is a degree of empirical disagreement regarding the range of spatial frequency information responsible for perceptual biases. Breaking continuous flash suppression (b. CFS) has explored these effects, showing similar biases for detecting fearful facial expressions. Recent findings from a b. CFS study highlight the role of high, rather than low spatial frequency content in determining faces' visibility. The present study contributes to ongoing discussions regarding the efficacy of b. CFS, and shows that the visibility of facial expressions vary according to how they are normalised for physical contrast and spatially filtered. Findings show that physical contrast normalisation facilitates fear's detectability under b. CFS more than when normalised for apparent contrast, and that this effect is most pronounced when faces are high frequency filtered. Moreover, normalising faces' perceived contrast does not guarantee equality between expressions' visibility under b. CFS. Findings have important implications for the use of contrast normalisation, particularly regarding the extent to which contrast normalisation facilitates fear bias effects.

Contrast normalisation masks natural expression-related differences and artificially enhances the perceived salience of fear expressions

Fearful facial expressions tend to be more salient than other expressions. This threat bias is to some extent driven by simple low-level image properties, rather than the high-level emotion interpretation of stimuli. It might be expected therefore that different expressions will, on average, have different physical contrasts. However, studies tend to normalise stimuli for RMS contrast, potentially removing a naturally-occurring difference in salience. We assessed whether images of faces differ in both physical and apparent contrast across expressions. We measured physical RMS contrast and the Fourier amplitude spectra of 5 emotional expressions prior to contrast normalisation. We also measured expression-related differences in perceived contrast. Fear expressions have a steeper Fourier amplitude slope compared to neutral and angry expressions, and consistently significantly lower contrast compared to other faces. This effect is more pronounced at higher spatial frequencies. With the exception of stimuli containing only low spatial frequencies, fear expressions appeared higher in contrast than a physically matched reference. These findings suggest that contrast normalisation artificially boosts the perceived salience of fear expressions; an effect that may account for perceptual biases observed for spatially filtered fear expressions.

Dynamic properties of internal noise probed by modulating binocular rivalry

Neural systems are inherently noisy, and this noise can affect our perception from moment to moment. This is particularly apparent in binocular rivalry, where perception of competing stimuli shown to the left and right eyes alternates over time. We modulated rivalling stimuli using dynamic sequences of external noise of various rates and amplitudes. We repeated each external noise sequence twice, and assessed the consistency of percepts across repetitions. External noise modulations of sufficiently high contrast increased consistency scores above baseline, and were most effective at 1/8Hz. A computational model of rivalry in which internal noise has a 1/f (pink) temporal amplitude spectrum, and a standard deviation of 16% contrast, provided the best account of our data. Our novel technique provides detailed estimates of the dynamic properties of internal noise during binocular rivalry, and by extension the stochastic processes that drive our perception and other types of spontaneous brain activity.

Although our perception of the world appears constant, sensory representations are variable because of the ‘noisy’ nature of biological neurons. Here we used a binocular rivalry paradigm, in which conflicting images are shown to the two eyes, to probe the properties of this internal variability. Using a novel paradigm in which the contrasts of rivalling stimuli are modulated by two independent external noise streams, we infer the amplitude and character of this internal noise. The temporal amplitude spectrum of the noise has a 1/f spectrum, similar to that of natural visual input, and consistent with the idea that the visual system evolved to match its environment.

Spatial summation of broadband contrast

Spatial summation of luminance contrast signals has historically been psychophysically measured with stimuli isolated in spatial frequency (i.e., narrowband). Here, we revisit the study of spatial summation with noise patterns that contain the naturalistic 1/fα distribution of contrast across spatial frequency. We measured amplitude spectrum slope (α) discrimination thresholds and verified if sensitivity to α improved according to stimulus size. Discrimination thresholds did decrease with an increase in stimulus size. These data were modeled with a summation model originally designed for narrowband stimuli (i.e., single detecting channel; Baker & Meese, 2011; Meese & Baker, 2011) that we modified to include summation across multiple-differently tuned-spatial frequency channels. To fit our data, contrast gain control weights had to be inversely related to spatial frequency (1/f); thus low spatial frequencies received significantly more divisive inhibition than higher spatial frequencies, which is a similar finding to previous models of broadband contrast perception (Haun & Essock, 2010; Haun & Peli, 2013). We found summation across spatial frequency channels to occur prior to summation across space, channel summation was near linear and summation across space was nonlinear. Our analysis demonstrates that classical psychophysical models can be adapted to computationally define visual mechanisms under broadband visual input, with the adapted models offering novel insight on the integration of signals across channels and space.

Increasing the Complexity of the Illumination May Reduce Gloss Constancy

We examined in which way gradual changes in the geometric structure of the illumination affect the perceived glossiness of a surface. The test stimuli were computer-generated three-dimensional scenes with a single test object that was illuminated by three point light sources, whose relative positions in space were systematically varied. In the first experiment, the subjects were asked to adjust the microscale smoothness of a match object illuminated by a single light source such that it has the same perceived glossiness as the test stimulus. We found that small changes in the structure of the light field can induce dramatic changes in perceived glossiness and that this effect is modulated by the microscale smoothness of the test object. The results of a second experiment indicate that the degree of overlap of nearby highlights plays a major role in this effect: Whenever the degree of overlap in a group of highlights is so large that they perceptually merge into a single highlight, the glossiness of the surface is systematically underestimated. In addition, we examined the predictability of the smoothness settings by a linear model that is based on a set of four different global image statistics.

Enhancing spatio-chromatic representation with more-than-three color coding for image description

The extraction of spatio-chromatic features from color images is usually performed independently on each color channel. Usual 3D color spaces, such as RGB, present a high inter-channel correlation for natural images. This correlation can be reduced using color-opponent representations, but the spatial structure of regions with small color differences is not fully captured in two generic Red-Green and Blue-Yellow channels. To overcome these problems, we propose new color coding that is adapted to the specific content of each image. Our proposal is based on two steps: (a) setting the number of channels to the number of distinctive colors we find in each image (avoiding the problem of channel correlation), and (b) building a channel representation that maximizes contrast differences within each color channel (avoiding the problem of low local contrast). We call this approach more-than-three color coding (MTT) to emphasize the fact that the number of channels is adapted to the image content. The higher the color complexity of an image, the more channels can be used to represent it. Here we select distinctive colors as the most predominant in the image, which we call color pivots, and we build the new color coding strategy using these color pivots as a basis. To evaluate the proposed approach, we measure the efficiency in an image categorization task. We show how a generic descriptor improves performance at the description level when applied to the MTT coding.

A normalized contrast-encoding model exhibits bright/dark asymmetries similar to early visual neurons

Biological sensory systems share a number of organizing principles. One such principle is the formation of parallel streams. In the visual system, information about bright and dark features is largely conveyed via two separate streams: the ON and OFF pathways. While brightness and darkness can be considered symmetric and opposite forms of visual contrast, the response properties of cells in the ON and OFF pathways are decidedly asymmetric. Here, we ask whether a simple contrast‐encoding model predicts asymmetries for brights and darks that are similar to the asymmetries found in the ON and OFF pathways. Importantly, this model does not include any explicit differences in how the visual system represents brights and darks, but it does include a common normalization mechanism. The phenomena captured by the model include (1) nonlinear contrast response functions, (2) greater nonlinearities in the responses to darks, and (3) larger responses to dark contrasts. We report a direct, quantitative comparison between these model predictions and previously published electrophysiological measurements from the retina and thalamus (guinea pig and cat, respectively). This work suggests that the simple computation of visual contrast may account for a range of early visual processing nonlinearities. Assessing explicit models of sensory representations is essential for understanding which features of neuronal activity these models can and cannot predict, and for investigating how early computations may reverberate through the sensory pathways.

Context-dependent lightness affects perceived contrast

Perceived contrast of a grating varies with its background (or mean) luminance: of the two gratings with the same photometric contrast the one on higher luminance background appears to have higher contrast. Does perceived contrast also vary with context-dependent background lightness even when the luminance remains constant? We investigated this question using a stimulus in which two equiluminant patches ("context squares", CSs) appear different in lightness. First we measured the lightness effect in a behavioral experiment. After ensuring that it was present for all participants, we conducted perceived contrast experiments, where participants judged the contrast of rectified incremental and decremental square-wave gratings superimposed on the CSs. For the incremental gratings participants' settings were significantly different for the two CSs. Specifically, perceived contrast was higher when the gratings were placed on the context square that was perceived lighter. In a follow-up experiment we measured perceived contrast of rectified gratings on isolated patches that differed in luminance. The pattern of results of the two experiments was consistent, demonstrating that possibly shared mechanisms underpin the effects of background luminance and context-dependent lightness on perceived contrast.

A cortical locus for anisotropic overlay suppression of stimuli presented at fixation

Human contrast sensitivity for narrowband Gabor targets is suppressed when superimposed on narrowband masks of the same spatial frequency and orientation (referred to as overlay suppression), with suppression being broadly tuned to orientation and spatial frequency. Numerous behavioral and neurophysiological experiments have suggested that overlay suppression originates from the initial lateral geniculate nucleus (LGN) inputs to V1, which is consistent with the broad tuning typically reported for overlay suppression. However, recent reports have shown narrowly tuned anisotropic overlay suppression when narrowband targets are masked by broadband noise. Consequently, researchers have argued for an additional form of overlay suppression that involves cortical contrast gain control processes. The current study sought to further explore this notion behaviorally using narrowband and broadband masks, along with a computational neural simulation of the hypothesized underlying gain control processes in cortex. Additionally, we employed transcranial direct current stimulation (tDCS) in order to test whether cortical processes are involved in driving narrowly tuned anisotropic suppression. The behavioral results yielded anisotropic overlay suppression for both broadband and narrowband masks and could be replicated with our computational neural simulation of anisotropic gain control. Further, the anisotropic form of overlay suppression could be directly modulated by tDCS, which would not be expected if the suppression was primarily subcortical in origin. Altogether, the results of the current study provide further evidence in support of an additional overlay suppression process that originates in cortex and show that this form of suppression is also observable with narrowband masks.

Sensitivity to gaze-contingent contrast increments in naturalistic movies: An exploratory report and model comparison

Sensitivity to luminance contrast is a prerequisite for all but the simplest visual systems. To examine contrast increment detection performance in a way that approximates the natural environmental input of the human visual system, we presented contrast increments gaze-contingently within naturalistic video freely viewed by observers. A band-limited contrast increment was applied to a local region of the video relative to the observer's current gaze point, and the observer made a forced-choice response to the location of the target (≈25,000 trials across five observers). We present exploratory analyses showing that performance improved as a function of the magnitude of the increment and depended on the direction of eye movements relative to the target location, the timing of eye movements relative to target presentation, and the spatiotemporal image structure at the target location. Contrast discrimination performance can be modeled by assuming that the underlying contrast response is an accelerating nonlinearity (arising from a nonlinear transducer or gain control). We implemented one such model and examined the posterior over model parameters, estimated using Markov-chain Monte Carlo methods. The parameters were poorly constrained by our data; parameters constrained using strong priors taken from previous research showed poor cross-validated prediction performance. Atheoretical logistic regression models were better constrained and provided similar prediction performance to the nonlinear transducer model. Finally, we explored the properties of an extended logistic regression that incorporates both eye movement and image content features. Models of contrast transduction may be better constrained by incorporating data from both artificial and natural contrast perception settings.