Model Of Human Visual Perception Research Articles

A deep new look at color.

Bowers et al. counter deep neural networks (DNNs) as good models of human visual perception. From our color perspective we feel their view is based on three misconceptions: A misrepresentation of the state-of-the-art of color perception; the type of model required to move the field forward; and the attribution of shortcomings to DNN research that are already being resolved.

Are Deep Neural Networks Adequate Behavioral Models of Human Visual Perception?

Deep neural networks (DNNs) are machine learning algorithms that have revolutionized computer vision due to their remarkable successes in tasks like object classification and segmentation. The success of DNNs as computer vision algorithms has led to the suggestion that DNNs may also be good models of human visual perception. In this article, we review evidence regarding current DNNs as adequate behavioral models of human core object recognition. To this end, we argue that it is important to distinguish between statistical tools and computational models and to understand model quality as a multidimensional concept in which clarity about modeling goals is key. Reviewing a large number of psychophysical and computational explorations of core object recognition performance in humans and DNNs, we argue that DNNs are highly valuable scientific tools but that, as of today, DNNs should only be regarded as promising-but not yet adequate-computational models of human core object recognition behavior. On the way, we dispel several myths surrounding DNNs in vision science.

JND-Guided Perceptually Color Image Watermarking in Spatial Domain

Watermarking is an effective solution for copyright protection and forensics tracking via hiding information into the image signal. Recently, spatial-embedding watermarking methods from different transform domain have been proposed rapidly and effectively protect the copyright of the color image. Compared to the individual frequency domain method, it has both advantages of spatial domain and frequency domain. Here, we proposed a novel spatial-embedding watermarking method based on an attended just noticeable difference (JND) model with color complexity, to achieve a good tradeoff between robustness and perceptual quality. In particular, at the spatial embedding level, the direct current (DC) coefficients are selected for embedding and the perceptual JND model is used to guide the amount of pixel modification to improve visual quality. Different from the previous JND model, we proposed an attended JND model that considers color complexity, which is more consistent of the human visual perception model. Compared with the other JND models, the proposed JND model is more suitable for the watermarking framework. Experimental results show that the proposed watermarking scheme has better performance than other existing watermarking schemes. This greatly benefits the practical implementations of the spatial-embedding watermarking methods.

Learning non-metric visual similarity for image retrieval

Measuring visual similarity between two or more instances within a data distribution is a fundamental task in image retrieval. Theoretically, non-metric distances are able to generate a more complex and accurate similarity model than metric distances, provided that the non-linear data distribution is precisely captured by the system. In this work, we explore neural networks models for learning a non-metric similarity function for instance search. We argue that non-metric similarity functions based on neural networks can build a better model of human visual perception than standard metric distances. As our proposed similarity function is differentiable, we explore a real end-to-end trainable approach for image retrieval, i.e. we learn the weights from the input image pixels to the final similarity score. Experimental evaluation shows that non-metric similarity networks are able to learn visual similarities between images and improve performance on top of state-of-the-art image representations, boosting results in standard image retrieval datasets with respect standard metric distances.

Using a model of human visual perception to improve deep learning

Deep learning algorithms achieve human-level (or better) performance on many tasks, but there still remain situations where humans learn better or faster. With regard to classification of images, we argue that some of those situations are because the human visual system represents information in a format that promotes good training and classification. To demonstrate this idea, we show how occluding objects can impair performance of a deep learning system that is trained to classify digits in the MNIST database. We describe a human inspired segmentation and interpolation algorithm that attempts to reconstruct occluded parts of an image, and we show that using this reconstruction algorithm to pre-process occluded images promotes training and classification performance.

Insight into color aesthetics from probabilistic perception modeling

AbstractA model of human visual perception is described. An observer is represented having no a‐priori preference for color or geometric properties of the objects in his visual field. The model includes computational representations of effectiveness and efficiency of visual memory allocation that are both based on likelihood. In the computations, the colorimetric quantities color difference, chroma, and lightness are of fundamental importance, being the state variables of model. Computer experiments are carried out verifying the validity of the model. Multiple corroborations between model outcome and commonly observed phenomena are pinpointed, indicating that aesthetical experience denotes an optimal state of visual system as to processing efficiency.

Real-time Colorized Video Images Optimization Method in Scotopic Vision

In low light environment, the surveillance video image has lower contrast ， less information and uneven brightness. To solve this problem, this paper puts forward a contrast resolution compensation algorithm based on human visual perception model. It extracts Y component from the YUV video image acquired by camera originally to subtract contrast feature parameters, then makes a proportional integral type contrast resolution compensation for low light pixels in Y component and makes index contrast resolution compensation for high light pixels adaptively to enhance brightness of the video image while maintains the U and V components. Then it compresses the video images and transmits them via internet. Finally, it decodes and displays the video image on the device of intelligent surveillance system. The experimental results show that, the algorithm can effectively improve the contrast resolution of the video image and maintain the color of video image well. It also can meet the real-time requirement of video monitoring.

Real-time Colorized Video Images Optimization Method in Scotopic Vision

<p><em>In low light environment, the surveillance video image has lower contrast</em><em>，</em><em>less information and uneven brightness. To solve this problem, this paper puts forward a contrast resolution compensation algorithm based on human visual perception model. It extracts Y component from the YUV video image acquired by camera originally to subtract contrast feature parameters, then makes a proportional integral type contrast resolution compensation for low light pixels in Y component and makes index contrast resolution compensation for high light pixels adaptively to enhance brightness of the video image while maintains the U and V components. Then it compresses the video images and transmits them via internet. Finally, it decodes and displays the video image on the device of intelligent surveillance system. The experimental results show that, the algorithm can effectively improve the contrast resolution of the video image and maintain the color of video image well. It also can meet the real-time requirement of video monitoring.</em><em></em></p>

Dynamic color image resolution compensation under low light

In low light environment, the dynamic color images are poorly identified. Therefore, this paper puts forward a kind of contrast resolution compensation algorithm. This algorithm is based on human visual perception model. Firstly, a color image is transformed from RGB color space into HSV color space, component of vector H remain unchanged; Secondly, extracting image feature parameters of vector V, then using contrast resolution compensate component of vector V in order to enhance image brightness; Thirdly, the component of vector S is linear stretched in order to recover the image color information; Finally, using the treated V elements, treated component of vector S and untreated component of vector H contracture a new enhanced image with RGB color space by inverse transform. Compared with histogram equalization and γ transformation algorithms, the compensation method can enhance images and improve image quality, and the compensated image has smaller color deviation. Using the quality assessment function of color image (CAF), the value of CAF with the compensated image is maximum, and the assessment result keeps consistent with the subjective assessment. This results show that the method is feasible with dynamic color image resolution compensation under low light.

Global interference during early visual processing: ERP evidence from a rapid global/local selective task

Visual perception depends on the integration of local elements of a visual scene into a global frame. Evidence from behavioral studies shows that (1) the detection of the global frame is faster than the detection of the local parts, a phenomenon called the global advantage, and that (2) an interference of the global shape is also present during local processing. Together, these effects are called the global precedence effect (GPE). Even if the global advantage appears to impact neural processing as early as the first 100 ms post-stimulus, previous studies failed to find a global interference effect before 200 ms post-stimulus. Using for the first time a rapid display of letter component stimuli during a global/local selective task in which conditions with perceptual conflict, congruent and incongruent conditions were considered, the present event-related potential (ERP) study shows a global interference effect occurring as early as the time range of the N1 component. In particular, only congruent stimuli elicited similar N1 amplitude during the global and local tasks, whereas an increased of the N1 amplitude during the global task was observed (as compared to the local task) for both stimuli with perceptual conflict and incongruent stimuli. This finding corroborates the recent neural models of human visual perception.

움직임 분석 기반의 시각인지 모델을 이용한 비디오 코딩 방법

본 논문에서는 인간 인지 기반 비디오 코딩을 위한 비디오 처리 방법을 개발한다. 제안하는 방법은 율-왜곡(rate-distortion) 최적화의 영향뿐만 아니라 제한적인 시, 공간 해상도, 지역적인 움직임 이력(history), visual saliency에 의한 인간 시각 인지를 고려한다. 이러한 인간의 인지적인 효과들을 고려하기 위하여 본 논문에서는 움직임 패턴을 모델링하고 Hedge 알고리듬을 사용하여 움직임 패턴을 결정하는 기법을 개발한다. 그 다음, 제안한 움직임 패턴과 기존의 visual saliency와의 결합을 통하여 인간 시각 인지 모델을 수립한다. 제안된 인간 시각 인지 모델을 구현하기 위하여 기존의 foveation filtering 방법을 확장한다. 시각적 자극이 덜한 지역만을 부드럽게(smoothing)하는 기존의 foveation filtering 기법과 비교하여 제안하는 foveation filtering 기법은 인간 시각 인지 모델에 따라 지역적으로 부드럽게 또는 지역적 특성을 향상시킴으로써, 시각적 자극이 덜한 지역에서 줄여진 대역폭을 효과적으로 시각적 자극이 큰 지역에서 사용하도록 이동 시킬 수 있는 장점이 있다. 제안된 방법의 성능은 전반적인 비디오 화질을 만족할 뿐만 아니라 인간이 인지하는 화질의 품질을 12%~44% 향상시킨다. We develop a video processing method that allows the more advanced human perception oriented video coding. The proposed method necessarily reflects all influences by the rate-distortion based optimization and the human visual perception that is affected by the visual saliency, the limited space-time resolution and the regional moving history. For reflecting the human perceptual effects, we devise an online moving pattern classifier using the Hedge algorithm. Then, we embed the existing visual saliency into the proposed moving patterns so as to establish a human visual perception model. In order to realize the proposed human visual perception model, we extend the conventional foveation filtering method. Compared to the conventional foveation filter only smoothing less stimulus video signals, the developed foveation filter can locally smooth and enhance signals according to the human visual perception without causing any artifacts. Due to signal enhancement, the developed foveation filter more efficiently transfers the bandwidth saved at smoothed signals to the enhanced signals. Performance evaluation verifies that the proposed video processing method satisfies the overall video quality, while improving the perceptual quality by 12%~44%.

Image Watermarking Based on Adaptive Models of Human Visual Perception

This paper proposes a digital image watermarking based on adaptive models of human visual perception. The algorithm exploits the local activities estimated from wavelet coefficients of each subband to adaptively control the luminance masking. The adaptive luminance is thus delicately combined with the contrast masking and edge detection and adopted as a visibility threshold. With the proposed combination of adaptive visual sensitivity parameters, the proposed perceptual model can be more appropriate to the different characteristics of various images. The weighting function is chosen such that the fidelity, imperceptibility and robustness could be preserved without making any perceptual difference to the image quality.

Bayesian decision theory as a model of human visual perception: Testing Bayesian transfer

Bayesian decision theory (BDT) is a mathematical framework that allows the experimenter to model ideal performance in a wide variety of visuomotor tasks. The experimenter can use BDT to compute benchmarks for ideal performance in such tasks and compare human performance to ideal. Recently, researchers have asked whether BDT can also be treated as a process model of visuomotor processing. It is unclear what sorts of experiments are appropriate to testing such claims and whether such claims are even meaningful. Any such claim presupposes that observers' performance is close to ideal, and typical experimental tests involve comparison of human performance to ideal. We argue that this experimental criterion, while necessary, is weak. We illustrate how to achieve near-optimal performance in combining perceptual cues with a process model bearing little resemblance to BDT. We then propose experimental criteria termed transfer criteria that constitute more powerful tests of BDT as a model of perception and action. We describe how recent work in motor control can be viewed as tests of transfer properties of BDT. The transfer properties discussed here comprise the beginning of an operationalization (Bridgman, 1927) of what it means to claim that perception is or is not Bayesian inference (Knill & Richards, 1996). They are particularly relevant to research concerning natural scenes since they assess the ability of the organism to rapidly adapt to novel tasks in familiar environments or carry out familiar tasks in novel environments without learning.

Visual perception model for architectural design

A model of human visual perception is presented. It is based on probability theoretic considerations. A probabilistic approach is suitable to model perception, as it can absorb the bio-induced complexity of the human vision process. In this way the perception and related concepts are quantified. The results provided by the model corroborate with common human vision experience confirming the validity of the model. The significance of distance on perception is described and exemplified in two applications for architectural design.

Discrimination Strategies of Humans and Rhesus Monkeys for Complex Visual Displays

By learning to discriminate among visual stimuli, human observers can become experts at specific visual tasks. The same is true for Rhesus monkeys, the major animal model of human visual perception. Here, we systematically compare how humans and monkeys solve a simple visual task. We trained humans and monkeys to discriminate between the members of small natural-image sets. We employed the "Bubbles" procedure to determine the stimulus features used by the observers. On average, monkeys used image features drawn from a diagnostic region covering about 7% +/- 2% of the images. Humans were able to use image features drawn from a much larger diagnostic region covering on average 51% +/- 4% of the images. Similarly for the two species, however, about 2% of the image needed to be visible within the diagnostic region on any individual trial for correct performance. We characterize the low-level image properties of the diagnostic regions and discuss individual differences among the monkeys. Our results reveal that monkeys base their behavior on confined image patches and essentially ignore a large fraction of the visual input, whereas humans are able to gather visual information with greater flexibility from large image regions.

JPEG quality transcoding using neural networks trained with a perceptual error measure.

A JPEG Quality Transcoder (JQT) converts a JPEG image file that was encoded with low image quality to a larger JPEG image file with reduced visual artifacts, without access to the original uncompressed image. In this article, we describe technology for JQT design that takes a pattern recognition approach to the problem, using a database of images to train statistical models of the artifacts introduced through JPEG compression. In the training procedure for these models, we use a model of human visual perception as an error measure. Our current prototype system removes 32.2% of the artifacts introduced by moderate compression, as measured on an independent test database of linearly coded images using a perceptual error metric. This improvement results in an average PSNR reduction of 0.634 dB.