Gersho's bounds on the asymptotic performance of vector quantizers are valid for vector distortions which are powers of the Euclidean norm. Yamada, Tazaki and Gray generalized the results to distortion measures that are increasing functions of the norm of their argument. In both cases, the distortion is uniquely determined by the vector quantization error, i.e., the Euclidean difference between the original vector and the codeword into which it is quantized. We generalize these asymptotic bounds to input-weighted quadratic distortion measures, a class of distortion measure often used for perceptually meaningful distortion. The generalization involves a more rigorous derivation of a fixed rate result of Gardner and Rao and a new result for variable rate codes. We also consider the problem of source mismatch, where the quantizer is designed using a probability density different from the true source density. The resulting asymptotic performance in terms of distortion increase in dB is shown...

By taking into account the properties and limitations of the human visual system (HVS), images can be more efficiently compressed, colors more accurately reproduced, prints better rendered, to mention a few major advantages. To achieve these goals it is necessary to build a computational model of the HVS. In this paper we give an introduction to the general issue of HVS-modeling and review the specific applications of visual quality assessment and HVS-based image compression, which are closely related. On one hand, these two examples demonstrate the common structure of HVS-models, on the other hand they also show how application-specific constraints influence model design. Recent vision models from these application areas are reviewed and summarized in a table for direct comparison.

We propose a new model for the prediction of distortion visibility in digital image sequences, which is aimed at use in digital video compression algorithms. The model is an extension of our spatial vision model with a spatio-temporal contrast sensitivity function and an eye movement estimation algorithm. Due to the importance of smooth pursuit eye movements when viewing image sequences, eye movements cannot be neglected in a spatio-temporal vision model. Although eye movements can be incorporated by motion compensation of the contrast sensitivity function, the requirements for this motion compensation are different than those for motion compensated prediction in video coding. We propose an algorithm for the estimation of smooth pursuit eye movements, under the worst-case assumption that the observer is capable of tracking all objects in the image. In image and image sequence compression, models which predict the visibility of coding distortions can be used to improve the visual quality of the total compression system. Many models have been proposed for the processing of spatial information in the human visual system to predict distortion visibility in coded images [1-5]. However, the number

While lossy image compression techniques are vital in reducing bandwidth and storage requirements, they result in distortions in compressed images. A reliable quality measure is a much needed tool for determining the type and amount of image distortion. The traditional subjective criteria, which involve human observers, are inconvenient, time-consuming, and influenced by environmental conditions. Widely used pixelwise measures such as the mean square error (MSE) cannot capture the artifacts like blurriness or blockiness, and do not correlate well with visual error perception. Attempts to improve quality measurement include incorporation of simple models of the human visual system (HVS) and multi-dimensional tool design. We review the criteria for monochrome compressed image quality from 1974 to 1999.

The important criteria used in subjective evaluation of distorted images include the amount of distortion, the type of distortion, and the distribution of error. An ideal image quality measure should therefore be able to mimic the human observer. We present a new image quality measure that can be used as a multidimensional or a scalar measure to predict the distortion introduced by a wide range of noise sources. Based on the Singular Value Decomposition, it reliably measures the distortion not only within a distortion type at different distortion levels but also across different distortion types. The measure was applied to Lena using six types of distortion (JPEG, JPEG 2000, Gaussian blur, Gaussian noise, sharpening and DC-shifting), each with five distortion levels.

This paper reviews the evolution of video quality measurement techniques and their current state of the art. We start with subjective experiments and then discuss the various types of objective metrics and their uses. We also introduce V-Factor, a “hybrid” metric using both transport- and bitstream information. Finally, we summarize the main standardization activities, such as the work of the Video Quality Experts Group (VQEG), and we take a look at emerging trends in quality measurement, including image preference, visual attention, and audiovisual quality.

This paper attempts to present a survey of quantitative quality/impairment metrics based on human visual sensitivity, and to describe a couple of quantitative metrics recently developed and reported for digital image and video coding.

The important criteria used in subjective evaluation of distorted images include the amount of distortion, the type of distortion, and the distribution of error. An ideal image quality measure should therefore be able to mimic the human observer. We present a new gray-scale image quality measure that can be used as a graphical or a scalar measure to predict the distortion introduced by a wide range of noise sources. Based on the Singular Value Decomposition, it reliably measures the distortion not only within a distortion type at different distortion levels but also across different distortion types. The measure was applied to five test images (Airplane, Boat, Goldhill, Lena, and Peppers) using six types of distortion (JPEG, JPEG 2000, Gaussian blur, Gaussian noise, sharpening, and DC-shifting), each with five distortion levels. Its performance is compared with PSNR and two recent measures.

Due to the improvement of image rendering processes, and the increasing importance of quantitative comparisons among synthetic color images, it is essential to define perceptually based metrics which enable to objectively assess the visual quality of digital simulations. In response to this need, this paper proposes a new methodology for the determination of an objective image quality metric, and gives an answer to this problem through three metrics. This methodology is based on the LLAB color space for perception of color in complex images, a recent modification of the CIELab1976 color space. The first metric proposed is a pixel by pixel metric which introduces a local distance map between two images. The second metric associates, to a pair of images, a global value. Finally, the third metric uses a recursive subdivision of the images to obtain an adaptative distance map, rougher but less expensive to compute than the first method.

Nonuniform quantizers for just not visible reconstruction errors in an adaptive intra-/interframe DPCM scheme for componentcoded color television signals are presented, both for conventional DPCM and for noise-shaping DPCM. Noise feedback filters that minimize the visibility of reconstruction errors by spectral shaping are designed for Y, R-Y, and B-Y. A dosed-form description of the "masking function" is derived which leads to the one-parameter "b quantizer" characteristic. Subjective tests that were carried out to determine visibility thresholds for reconstruction errors for conventional DPCM and for noise shaping DPCM show significant gains b noise shaping. For a transmission rate of around 30 Mbits/s, reconstruction errors are almost always below the visibility threshold if variable length encoding of the prediction error is combined with noise shaping within a 3:1:1 system.