In this paper we develop a computational model of visual masking based on psychophysical data. The model predicts how the presence of one visual pattern affects the detectability of another. The model allows us to choose texture patterns for computer graphics images that hide the effects of faceting, banding, aliasing, noise and other visual artifacts produced by sources of error in graphics algorithms. We demonstrate the utility of the model by choosing a texture pattern to mask faceting artifacts caused by polygonal tesselation of a flat-shaded curved surface. The model predicts how changes in the contrast, spatial frequency, and orientation of the texture pattern, or changes in the tesselation of the surface will alter the masking effect. The model is general and has uses in geometric modeling, realistic image synthesis, scientific visualization, image compression, and image-based rendering.

Lossy compression algorithms used in digital video systems produce artifacts whose visibility strongly depends on the actual image content. Simple error measures such as RMSE or PSNR, albeit popular, ignore this important fact and are only a mediocre predictor of perceived quality. Many applications require more reliable assessment methods. This paper discusses issues in vision modeling for perceptual video quality assessment (PVQA). Its purpose is not to describe a particular model or system, but rather to summarize and to provide pointers to up-to-date knowledge of important characteristics of the human visual system, to explain how these characteristics may be incorporated in vision models for PVQA, to give a brief overview of the state-of-the-art and current efforts in this field, and to outline directions for future research.

A letter in the peripheral visual field is much harder to identify in the presence of nearby letters. This is called "crowding". In general, masking is a procedure: introducing any "mask" pattern that affects discriminability of the signal. Crowding conforms to the masking paradigm, but the crowding effect is unlike ordinary masking. Here we characterize crowding, and present diagnostic tests that distinguish it from ordinary masking. In ordinary masking, the signal disappears. In crowding, it remains visible, but is ambiguous, confounded with its neighbors. Masks are usually effective only if they overlap the signal, but the crowding effect extends over a large region. The width of that region is proportional to signal eccentricity from the fovea and independent of signal size, mask size, signal and mask font, and number of masks. At 4 deg eccentricity, the threshold contrast for identification of a 0.32 deg signal letter is elevated (up to six-fold) by mask letters anywhere in a 2.3 deg region, seven times wider than the signal. In ordinary masking, threshold contrast rises as a power function of mask contrast, with a shallow log-log slope of 0.5 to 1, while in crowding, threshold is a sigmoidal function of mask contrast, with a steep log-log slope of 2 at close spacing. Most remarkably, although the threshold elevation decreases exponentially with spacing, the threshold and saturation contrasts of crowding are independent of spacing. Finally, ordinary masking is similar for detection and identification, but crowding occurs only for identification, not detection. More precisely, crowding occurs only in tasks that cannot be done based on a single detection by coarsely coded feature detectors. These results (and observers' introspections) suggest that ordinary masking b...

An algorithm for the detection of visually relevant luminance features is presented. The algorithm is motivated and directed by current models of the visual system. The algorithm detects edges (sharp luminance transitions) and narrow bars (luminance cusps) and marks them with the proper polarity. The image is first bandpass filtered with oriented filters at a number of scales an octave apart. The suprathreshold image contrast details at each scale are then identified and are compared across scales to find locations in which the signal polarity (sign) is identical at all scales, representing a minimal level of phase congruence across scales. These locations maintain the polarity of the bandpass-filtered image. The result is a polarity-preserving features map representing the edges with pairs of light and dark lines or curves on corresponding sides of the contour. Similarly, bar features are detected and represented with single curves of the proper polarity. The algorithm is implemented without free (fitted) parameters. All parameters are directly derived from visual models and from measurements on human observers. The algorithm is shown to be robust with respect to variations in filter parameters and requires no use of quadrature filters or Hilbert transforms. The possible utility of such an algorithm within the visual system and in computer vision applications is discussed. Keywords—Biological systems, edge detection, image matching, image processing, machine vision. I.

We studied spatial integration at low contrasts by testing the detection thresholds of multi Gabor element displays, examining configuration parameters such as orientation uniformity, contour smoothness, continuity, spacing and relative phase. We find that detectability depends on stimulus geometry and is constrained by collinearity and proximity spatial relations. For a "coherent" configuration (e.g. smooth and continuous contour) thresholds decrease linearly with increased number of elements, on a log-log scale, as expected from probability summation (q = 4). However, for a "non-coherent" configuration (e.g. spacing ? 4 or jagged contour) thresholds are little affected by the number of patches. Similar behavior is observed for supra-threshold stimuli embedded in bandpass noise. These results suggest that contrast integration is primarily based on local mechanisms and constrained by contour properties. These local mechanism are possibly mediated by lateral interactions in the primary ...

We report contrast detection, contrast increment, contrast masking, orientation discrimination, and spatial frequency discrimination thresholds for spatially localized stimuli at 4 ° of eccentricity. Our stimulus geometry emphasizes interactions among overlapping visual filters and differs from that used in previous threshold measurements, which also admits interactions among distant filters. We quantitatively account for all measurements by simulating a small population of overlapping visual filters interacting through divisive inhibition. We depart from previous models of this kind in the parameters of divisive inhibition and in using a statistically efficient decision stage based on Fisher information. The success of this unified account suggests that, contrary to Bowne [Vision Res. 30, 449 (1990)], spatial vision thresholds reflect a single level of processing, perhaps as early as primary visual cortex. © 2000 Optical Society of America [S0740-3232(00)02311-5] OCIS codes: 330.0330, 330.1800, 330.4060, 330.5510, 330.6100, 330.7310. 1.

Our objective is to nd a theoretical watermarking capac-ity bound of digital images based on domain-speci c masking e ects. In this paper, we rst show the capacity of private wa-termarking in which the power constraints are not uniform. Then, we apply several domain-speci c Human Vision Sys-tem approximation models to estimate the power constraints and then show the theoretical watermarking capacity of an image in a general noisy environment. Note that we con-sider all pixels, watermarks and noises are discrete values, which occur in realistic cases. 1.

Multimedia authentication techniques are required in order to ensure trustworthiness of multimedia data. They are usually designed based on two kinds of tools: digital signature or watermarking. Digital signature is a non-repudiatible, encrypted version of the message digest extracted from the data. Watermarking techniques consider multimedia data as a communication channel transmitting owner identification or content integrity information. Given an objective for multimedia authentication to reject the crop-and-replacement process and accept content-preserving manipulations, traditional digital signature or watermarking methods cannot be directly applied. In this thesis, we first propose robust digital signature methods that have proved to be useful for such types of content authentication. Also, we have developed a novel semi-fragile watermarking technique to embed the proposed robust digital signatures. We have implemented a unique Self-Authentication-and-Recovery Images (SARI) system, which can accept quantization-based lossy compression to a determined degree without any false alarms and can sensitively detect and locate malicious manipulations. Furthermore, the corrupted areas can be approximately recovered by the information hidden in the other part of the content. The amount of information embedded in our SARI system has nearly reached the theoretical maximum zero-error information hiding capacity of digital images. Watermarking is a promising solution that can protect the copyright of multimedia data through transcoding. A reasonable expectation of applying watermarking techniques for copyright protection is to consider specific application scenarios, because the distortion behavior involved in these cases (geometric distortion and pixel value distortion) could be reasonably predictable. We propose a practical public watermarking algorithm that is robust to rotation, scaling, and/or translation (RST) distortion. This proposed algorithm plays an important role in our design of the public watermarking technique which survives the image print-and-scan process. In addition, we present our original work in analyzing the theoretical watermarking capacity bounds for digital images, based on the information theory and the characteristics of the human vision system. We investigate watermarking capacity in three directions: the zero-error capacity for public watermarking in magnitude-bounded noisy environments, the watermarking capacity based on domain-specific masking effects, and the watermarking capacity issues based on sophisticated Human Vision System models.

The traditional mean-squared-error (MSE) or peak-signal-to-noise-ratio (PSNR) error measures are mainly focused on the pixel-by-pixel difference between the original and compressed images. Such metrics are improper for subjective quality or fidelity assessment, since human perception is very sensitive to correlations between adjacent pixels. In this work, we explore the Haar wavelet to model the space-frequency localization property of human visual system (HVS). It is shown that the physical contrast in different resolutions can be easily represented in terms of transform coefficients. We model HVS with the Haar filter with several visual mechanisms and develop a subjective quality measure which is more consistent with human observation experience. Keywords: image quality assessment, compression artifact measure, human visual system model (HVS), Haar wavelet. 1. INTRODUCTION Image quality assessment is an important task for evaluation of effectiveness and performance of various loss...

This paper involves designing, implementing, and testing of a locally adaptive perceptual masking threshold model for image compression. This model computes, based on the contents of the original images, the maximum amount of noise energy that can be injected at each transform coefficient that results in perceptually distortion-free still images or sequences of images. The adaptive perceptual masking threshold model can be used as a pre-processor to a JPEG compression standard image coder. DCT coefficients less than their corresponding perceptual thresholds can be set to zero before the normal JPEG quantization and Huffman coding steps. The result is an image-dependent gain in the bit rate needed for transparent coding. In an informal subjective test involving 318 still images in the AT&T Bell Laboratory image database, this model provided a gain in bit-rate saving on the order of 10 to 30 %. 1. INTRODUCTION In recent years, JPEG has emerged as a standard method of image compression [...