Showing papers on "Human visual system model published in 1991"

Fast Perceptual Learning in Visual Hyperacuity

Abstract: In many different spatial discrimination tasks, the human visual system exhibits hyperacuity-level performance by evaluating spatial relations with the precision of a fraction of a photoreceptor''s diameter. We propose that this impressive performance depends in part on a fast learning process that uses relatively few examples and occurs at an early processing stage in the visual pathway. We demonstrate that it is possible to synthesize from a small number of examples a simple (HyperBF) network that attains the required performance level. We verify with psychophysical experiments some key predictions of our conjecture. We show that fast stimulus-specific learning indeed takes place in the human visual system and that this learning does not transfer between two slightly different hyperacuity tasks.

The Plenoptic Function and the Elements of Early Vision

Abstract: This chapter contains sections titled: The Plenoptic Function, Plenoptic Structures, The Plenoptic Function and Elemental Measurements in Early Vision, Plenoptic Measurements in the Human Visual System, Periodic Tables for Early Vision, Further Computations, Conclusion, Appendix, References

A regularized iterative image restoration algorithm

Abstract: The development of the algorithm is based on a set theoretic approach to regularization. Deterministic and/or statistical information about the undistorted image and statistical information about the noise are directly incorporated into the iterative procedure. The restored image is the center of an ellipsoid bounding the intersection of two ellipsoids. The proposed algorithm, which has the constrained least squares algorithm as a special case, is extended into an adaptive iterative restoration algorithm. The spatial adaptivity is introduced to incorporate properties of the human visual system. Convergence of the proposed iterative algorithms is established. For the experimental results which are shown, the adaptively restored images have better quality than the nonadaptively restored ones based on visual observations and on an objective criterion of merit which accounts for the noise masking property of the visual system. >

The how and why of what went where in apparent motion: Modeling solutions to the motion correspondence problem.

Abstract: A model that is capable of maintaining the identities of individuated elements as they move is described. It solves a particular problem of underdetermination, the motion correspondence problem, by simultaneously applying 3 constraints: the nearest neighbor principle, the relative velocity principle, and the element integrity principle. The model generates the same correspondence solutions as does the human visual system for a variety of displays, and many of its properties are consistent with what is known about the physiological mechanisms underlying human motion perception. The model can also be viewed as a proposal of how the identities of attentional tags are maintained by visual cognition, and thus it can be differentiated from a system that serves merely to detect movement.

Relative visual performance: A basis for application:

Abstract: Visual performance is defined in terms of the speed and accuracy of processing visual information. To evaluate illuminated tasks in terms of visual performance, it is necessary to have a valid computational model that relates measurable, salient aspects of the visual environment (e.g. target contrast) to measurable human responses (e.g. visual response time). A model of visual performance should be independent of the influence of non-visual factors as they influence speed and accuracy. It is also important to have practical application tools that can measure the salient aspects of the visual environment and compute (predict) visual performance for real tasks. This paper describes the basis for a visual performance model and how it can be applied using a computer imaging device.

New approach to palette selection for color images

Abstract: We apply the vector quantization algorithm proposed by Equitz to the problem of efficiently selecting colors for a limited image palette. The algorithm performs the quantization by merging pairwise nearest neighbor (PNN) clusters. Computational efficiency is achieved by using k- dimensional trees to perform fast PNN searches. In order to reduce the number of initial image colors, we first pass the image through a variable-size cubical quantizer. The centroids of colors that fall in each cell are then used as sample vectors for the merging algorithm. Tremendous computational savings is achieved from this initial step with very little loss in visual quality. To account for the high sensitivity of the human visual system to quantization errors in smoothly varying regions of an image, we incorporate activity measures both at the initial quantization step and at the merging step so that quantization is fine in smooth regions and coarse in active regions. The resulting images are of high visual quality. The computation times are substantially smaller than that of the iterative Lloyd-Max algorithm and are comparable to a binary splitting algorithm recently proposed by Bouman and Orchard.

Selective attention modulates extrastriate visual regions in humans during visual feature discrimination and recognition

Abstract: Positron emission tomography (PET) was used to identify regions of the human visual system which were selectively modulated by attention during feature discrimination and recognition tasks. In a first experiment, subjects were cued to the shape, colour or speed of visual stimulus arrays during a same-different match-to-sample paradigm. The psychophysical sensitivity for discriminating subtle attribute variations was enhanced by selective attention. Correspondingly, the neural activity (as measured by blood flow changes) in different visual associative regions was enhanced when subjects attended to different attributes of the same stimulus (intraparietal sulcus for speed; collateral sulcus and dorsolateral occipital cortex for colour; collateral sulcus, fusiform and parahippocampal gyri, superior temporal sulcus for shape). These regions appeared to be specialized for processing the selected attribute. Attention to a visual feature, therefore, enhances the psychophysical sensitivity as well as the neural activity of specialized processing regions of the human visual system. In a second experiment the effect of target probability (which biases attentional selection) was studied during visual search tasks involving the recognition of a single-feature (i.e. colour) or a feature-conjunction (i.e. colour x orientation) target. Target probability positively modulated neural activity of extrastriate visual regions, which were related to the single-feature or feature-conjunction processing level. These results suggest that selective attention can influence different processing levels in the visual system, possibly reflecting a facilitatory effect on different visual computations or task components.

Evaluating the EM algorithm for image processing using a human visual fidelity criterion

Abstract: A new image quality metric consistent with the properties of the human visual system is derived. Using the EM algorithm, the authors restore a blurred image and quantify the improvement in image quality with both the new metric and the mean square error (MSE). From these results, the advantages of the new metric are obvious. The EM algorithm is modified according to the underlying mathematical structure of the new metric, which results in improved performance. >

Testing and treatment of visual dysfunctions

Abstract: A method for testing and treating visual dysfunctions (for example amblyopia) includes steps of providing an interactive visual game that presents to the patient a visual task, the visual game employing images scaled to the threshold visual parameter or parameters of the patient, and presenting a nonvisual reward to the patient for successful completion of the task. Also, apparatus for treating visual dysfunctions in a patient includes means for presenting an image to the patient, the image presenting a visual task to the patient, means for scaling the image to about the patient's threshold value for a visual parameter or parameters, means for providing an interaction between the patient and the image, and means for providing a reward for successful completion of the task. Also, an interactive video apparatus includes video display means for simultaneously stimulating a subject's interest and stimulating the subject's vision at about the maximum value of a visual parameter or parameters of which the subject is capable, means for receiving input from the subject, and means for providing to the subject a nonvisual reward for subject input that is appropriate to an image presented to the subject on the video display means. Also, a computerized method for testing a visual parameter or parameters comprises providing an interactive visual test that presents to the patient an image to be identified and, in step increments of increasing difficulty of the parameter in question, determining that patient's threshold value for the visual parameter or parameters by providing for interaction between the patient and the image.

Eye monitor: microcomputer-based instrument uses an internal mode to track the eye

Abstract: An eye monitor whose design is inspired by the human visual system is presented. The monitor incorporates an internal representation or model of what the eye looks like to a video camera. The system can measure the position of the eyes and the size of the pupil in the presence of interfering noise and in patients wearing eyeglasses or contact lenses, and it tolerates defocusing due to small movements in depth by the patient. The system makes real-time correction for head and eye movements while measuring pupillary responses to controlled light stimuli. The design and software and hardware components are described, and some applications are noted. Its use for early clinical detection of visual diseases by objectively measuring pupillary responses to carefully controlled light stimuli is examined as an example. Some general observations about using computers in medical measurements are made. >

VISIT: A Neural Model of Covert Visual Attention

Abstract: Visual attention is the ability to dynamically restrict processing to a subset of the visual field. Researchers have long argued that such a mechanism is necessary to efficiently perform many intermediate level visual tasks. This paper describes VISIT, a novel neural network model of visual attention. The current system models the search for target objects in scenes containing multiple distractors. This is a natural task for people, it is studied extensively by psychologists, and it requires attention. The network's behavior closely matches the known psychophysical data on visual search and visual attention. VISIT also matches much of the physiological data on attention and provides a novel view of the functionality of a number of visual areas. This paper concentrates on the biological plausibility of the model and its relationship to the primary visual cortex, pulvinar, superior colliculus and posterior parietal areas.

Removal of subjective redundancy from DCT-coded images

Abstract: The removal of subjective redundancy from video images has recently become an important area of study. A suggested method of removing this redundancy from transform-coded images is through the psychovisual thresholding and quantisation of the image transform coefficients. In this paper, the coefficient thresholding and quantisation levels are based on the combined effects of spatial masking and the varying sensitivity of the human visual system to different spatial frequencies and levels of luminance. By combining the Discrete Cosine Transform (DCT) method of image coding with psychovisual thresholding and quantisation schemes, subdistortion motion video bit-rates as low as 2.5 Mbit/s (non-interlaced 25 frame-per-second video) have been obtained without the need for interframe coding.

Compression of stereo images using subsampling and transform coding

Abstract: The proposed approach to stereo image coding takes advantage of the singleness of vision property of the human visual system. A stereo image pair, in which one of the images is low-pass filtered and subsampled, is perceived stereoscopically as a sharp 3-D image. The depth information is perceived due to the stereopsis effect, and the sharpness is maintained due to the details in the nonfiltered image. Low-pass filtering, subsampling, and discrete cosine transform image coding are used for the compression of the stereo images. A methodology for the evaluation of the compression effects on the 3-D perception of compressed stereo images is presented. It is based on measurements of response time and accuracy of human subjects performing simple 3-D perception tasks.

Fundamental properties of medical image perception.

Abstract: With a mind toward the effective acquisition, processing, presentation, and reading of radiological images, a survey of how the human visual system perceives images is presented here. The level is chosen to be suitable for the radiologist, and the relative emphasis on the various visual cues of luminance, color, form, texture, motion, and depth is chosen based on their importance with radiological images. Examples of the radiological relevance of the various visual properties are given. We cover first what the visual system's behavior is and then survey some of the properties of the physiological mechanisms that provide this behavior.

Dct transform compression, transmission and recovery of digital color using virtual filtering mechanism

Abstract: A color imagery compression/decompression mechanism employs a contrast sensitivity function model of the human visual system for color images, to derive normalization values for compressing chromatic components of the color imagery data. The CSF model is defined in terms of an ADT color space transform (15), the chromatic channels of which conform with the band-limited sensitivity of the visual system. The transform for obtaining the ADT components is executable by reduced complexity logic operators. Each of the ADT components is block transform coded using DCT transform coefficients (18) that are normalized (24) in accordance with a `weighted` (46, 45) CSF model (42) of the human visual system, which effectively performs a low pass filtering of the chromatic (DT) channels. This low pass filter-weighting of the normalization values in the compression, transmit end (10) of the system produces a set of normalized coefficients which effectively blur the image and reduce the entropy in the chromatic channels and may be implemented using a Gaussian function. In the decompression process at the receiver (14), a corresponding set of denormalization coefficients (32), which are generated in accordance with the same chromatic CSF model of the human visual system, are not weighted, so that the cascading of the normalization and denormalizing operators in the transmitter and receiver yields a product less than unity, simulating the entropy-reducing effect of subsampling.

Cognitive view mechanism for multimedia database system

Abstract: Describes a cognitive human interface for visual interaction with image database systems. The authors' approach gives a general framework of visual interaction. They adopt both an image model and a user model to interpret and operate the contents of image data from the user's viewpoint. The image model describes the physical constraints of image data, while the user model reflects the visual perception processes of the user. They propose the algorithms for typical visual interaction styles; a query by visual example (QVE) and a query by subjective descriptions (QBD). The former includes a sketch retrieval function, and a similarity retrieval function and the latter includes a sense retrieval function. These algorithms are developed for their experimental database system, the TRADEMARK and the ART MUSEUM. These functions use a pictorial index created by image analysis and a personal index automatically learned as the user model. >

Perception of 3-D Illusory Surface with Binocular Viewing

Abstract: A new discovery was made concerning the illusion of 3-D surface perception in binocular fusion. From the visual stimuli of disparity given only partially along the contour of an object, the human visual system can perceive the 3-D surface (not only planar but also curved) where there are no visual stimuli to provide any depth information. There are several new findings concerning this illusion: the interaction between several illusory surfaces can be observed; an edge line can be perceived as an intersection of two illusory surfaces; the occluding effect between two illusory surfaces can be recognized; the transparent effect of the illusory surface can be seen with an adequate visual stimulus. These newly found illusory phenomena have close relations with the visual functions of 3-D space perception.

Characterizing visual languages

Abstract: A better understanding of the visual character of languages is important in developing one's ability to exploit the human visual system. The author briefly outlines Goodman's (1976) distinction between notational and analog languages, and describes its use in developing the notion of syntactic and semantic density as the defining characteristic of visual languages. Several languages are evaluated for their use of density. He concludes that practical languages are most visually effective when their layout is constrained by an important semantic domain. >

Lossy image compression for digital medical imaging systems

Abstract: Image compression at rates oflO:1 orgreatercould make picture archiving and communication systems (PACS) much more responsive and economically attractive. A protocol is described for subjective and objective evaluation of the fidelity of compressed/decompressed images compared to originals. The results of its application to four representative and promising compression methods are presented. The four compression methods examined are predictive pruned tree-structured vector quantization, fractal compression, the full-frame discrete cosine transform with equal weighting of block bit allocation, and the full-frame discrete cosine transform with human visual system weighting of block bit allocation. A protocol was developed for side-by-side observer comparison of reconstructed images with originals. Three 1024 x 1024 computed radiography (CR) images andtwo 512 x 512 x-ray computed tomography(CT) images were viewed at six bit rates by nine radiologists at the University of Washington Medical Center. The radiologists' subjective evaluations of image fidelity were compared to calculations of mean square error for each decompressed image.

A contrast definition for logarithmic images in the continuous setting

Abstract: The specific laws and structures adapted to the processing and the analysis of logarithmic images, such as the images obtained by transmitted light or those outcoming from the human visual system, have been defined in the setting of the Logarithmic Image Processing Model (Jourlin,Pin0li, 1987 and l988).This model called L.I.P. also allowed the introduction of a natural contrast definition, which is closely linked with logarithmic images and setted up in an vectorial structure. However, only the discrete case has been good studied until now (Jourlin, Pinoli, Zeboudj, 1989). With the tools developed elsewhere (Pinoli, 1987a and 1987b), this paper extends the contrast notion to the continuous case, as it was previously announced.

Support provided with a security element

Abstract: A support is provided with a machine-detectable security element, particularly a copying-security element. The security element comprises a first image perceptible to the human visual system, said first image incorporating a second image that is substantially imperceptible to the human visual system.

Coding-oriented segmentation based on Gibbs-Markov random fields and human visual system knowledge

Abstract: A new segmentation algorithm for still black and white images is introduced. This algorithm forms the basis of a region-oriented sequence coding technique, currently under development. The algorithm models the human mechanism of selecting regions both by their interior characteristics and their boundaries. This is carried out in two different stages: with a preprocessing that takes into account only gray level information, and with a stochastic model for segmented images that uses both region interior and boundary information. In the stochastic model, the gray level information within the regions is modeled by stationary Gaussian processes, and the boundary information by a Gibbs-Markov random field (GMRF). The segmentation is carried out by finding the most likely realization of the joint process (maximum a posteriori criterion), given the preprocessed image. For decreasing the computational load while avoiding local maxima in the probability function, suboptimal versions of the algorithm are proposed. >

Image quality metrics for volumetric laser displays

Abstract: IMAGE QUALITY METRICS FOR VOLUMETRIC LASER DISPLAYSRodney Don Williams and Daniel DonohooTexas Instruments IncorporatedComputer Systems LaboratoryUser Systems Engineering LaboratoryDallas, Texas, USAABSTRACTThis paper addresses the extensions to the image quality metrics and related human factors research that are needed to establish thebaseline standards for emerging volume display technologies. The existing and recently developed technologies for multiplanarvolume displays are reviewed with an emphasis on basic human visual issues. Human factors image quality metrics and guidelinesareneeded to firmly establish this technology inthe marketplace. Thehuman visualrequirements and the display designtradeoffs forthese prototype laser-based volume displays are addressed and several criticalimage quality issues identified for further research. TheAmerican National Standard for Human Factors Engineering ofVisual Display Terminal Workstations (ANSI HFS-100) and otherintemationalstandards (ISO, DIN) can serve as a starting point, butthis researchbase mustbe extended to providenew image qualitymetrics for this new technology for volume displays.1. INTRODUCTIONThis paper provides an overview of volumetric display technology and presents the human factors research needed for effectivevolumetric information displays. Texas Instruments has developed an autostereoscopic multiplanar volume display technology overthe last 4 ars' During the development of this technology, we conducted an extensive review of previous technologies for truevolume displays and identified numerous research issues associated with volume technology. The American National Standard forHuman Factors Engineering of Visual Display Terminal Workstations can serve as a starting point, but this research base must beextended to provide the image quality metrics for volume displays.or as a physical 3-D volume as described herein. For some tasks, the parallax cues from stereoscopic presentations are sufficient.However, for many multiple-person (team) tasks where numerous views are required, a true volume display is preferred. This articlereviews the various technologies for implementing these volume displays and the image quality issues associated withthis emergingtechnology.In this context, the term "volumetric display" refers to a display device that uses a physical volume to present images having height,depth, and width. These images occupy physical space and are not a 2-D rendering of a 3-D image. A "voxel" is a volume displaypicture element(pixel)having a physical X, Y, and Z dimensionality in a specific location in a volume display. Note that these termsare often used to describe a virtual 2-D display with additional 3-D cues such as parallax, perspective, etc. However, for the purposeof this paper, I am referring to an actual physical volume of three dimensions.While the human visual system was created to effectively perceive and comprehend the world in a full three dimensions, currentcomputer graphics displays provide only 2-D views of the real world. This incompatibility becomes most pronounced when a useris asked to monitor and control the relative position of objects in a volume of space. We need information displays that bettercommunicate the relative position of 3-D objects in real space and provide complete visual cues.Current 2-D perspective displays have limitations when used for dynamic volumetric tasks such as battle management and air

Computer vision sensing of stress cracks in corn kernels

Abstract: Maintaining high quality of com is important to both com producers and buyers. Stress crack detection remains one of the most important tasks in com quality inspection. Such a measure of quality would be helpful in assessing not only the end-use values of the com, but also the drying method used and the amount of expected breakage due to subsequent handling procedures. A computer vision system was developed for automatic detection of corn stress cracks which simulates the processes that the human visual system uses to perceive the stress cracks from the com kernel in the conventional candling method. The stress crack detection system consisted of four consecutive stages; windowing, edge detection, feature representation, and classification. A set of performance criteria was developed to evaluate the stress crack detection system and used to compare the performance of different configurations on several com varieties. Evaluation results showed that the system configured with the circular band operator, the Duda road operator, and the Hough transform performed best; with success rates of 78.2% and failure rates of 8.2% of the classification made by one human expert. The performance measures of the system with this configuration were equal or superior to that of other human inspectors. The accuracy of the system was 91.8% when the system was used to distinguish only stress cracked-kemels from sound kernels.

Hybrid image coding scheme in incorporating human visual system characteristics

Abstract: In this paper, a hybrid image coding scheme combining the advantages of both classified vector quantization (CVQ) and discrete cosine transform (DCT) is proposed. An image vector is classified according to its content activity. Vectors of low activity (shade vectors) are coded using CVQ. For vectors of high activity (edge vectors), DCT is used. To enhance the visual quality of the decoded image, the modulation transfer function (MTF) of the human visual system (HVS) model is incorporated into the DCT coding algorithm. Simulation results showed that the proposed approach allowed good subjective image quality to be obtained at relatively low bit rates. The improvement in quality is even more pronounced at low bit rates, if this

Variable bit-rate coding based on human visual system

Abstract: One of the merits with the introduction of variable bit-rate coding to moving pictures is the constancy of picture quality. However, considering the fact that conventional SNR cannot reflect the result of subjective assessment sufficiently, other quality measures should be considered to design the variable bit-rate coders. In this paper, weighted SNR (WSNR), which is calculated by weighting coding errors with the spatial frequency response of human vision, is introduced as a quality measure in place of SNR, and the variable bit-rate coder based on this measure is considered. Then, subjective assessment is carried out and the effectiveness of our approach is confirmed.

Formation of Image Objects in Human Vision

Abstract: The stage of human vision in which objects are formed apparently precedes the final determination of many object properties: object identity, brightness, size, and location. Analogously in computer vision, this stage must precede isolated display and measurement of the object. This paper brings together several ideas about how this grouping of space into objects may be done by the human visual system, for example when it is viewing medical images. The result of this synthesis is a new model of form perception.