Browse Prior Art Database

Objective Image and Video Quality Assessment

IP.com Disclosure Number: IPCOM000015613D
Original Publication Date: 2002-Jan-07
Included in the Prior Art Database: 2003-Jun-20
Document File: 5 page(s) / 133K

Publishing Venue

IBM

Abstract

Objective Video Quality Assessment

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 29% of the total text.

Page 1 of 5

Objective Image and Video Quality Assessment

Objective Video Quality Assessment

I. INTRODUCTION

  Perceptual video quality is often essential in many video applications. For examples, video restoration and enhancement try to optimize certain quality metrics. In video coding and communications, a good quality metrics is fundamentally important throughout the system. At the encoder, it is used in optimal coding and pre-processing algorithms. At the decoder, it can be used in post-processing to improve the visual quality. Quality metrics is also very instrumental for video transcoders, network servers and switches to control and monitor the video service quality.

  The existing international standard for subjective video quality assessment relies on human experts. Indeed the industry has used this standard for many years. However, it is inconvenient and expensive. The goal of objective video quality assessment research is to develop some quality metrics that can be implemented by mathematical algorithm to automatically predict the perceptual quality of video. The most commonly used objective quality metrics is Peak Signal-to-Noise ratio (PSNR). However, PSNR is widely known to have poor correlation with perceptual image and video quality.

  Most recent research work on objective video quality measurement is conducted by the members of Video Quality Experts Group (VQEG)[4]. VQEG was formed in 1997 to develop and standardize method for video objective quality measurement. Many members are also active in the international telecommunications union (ITU). A major goal of VQEG is to develop and evaluate full reference (FR) video quality assessment methods leading to an ITU standard. VQEG has completed Phase I test. 10 proposals were evaluated, including PSNR and the well-known Sarnoff and NASA models. However, no proposal statistically outperformed PSNR in the evaluation and consequently no recommendation was made for ITU standard. VQEG is now preparing for the next round of test.

II. TRADITIONAL METHODS

The most widely used objective measurement method PSNR is defined as:

L2210)110N== -

k

PSNR

log

(

xk y

N

k

1

Page 2 of 5

DCTune models.

  In general, the error-sensitivity based methods can be summarized as the follows, though variances exist. First, the distorted image or video signal is subtracted from the original signal and the error signal is subject to channel decomposition. There are many variations for channel decomposition, such as wavelet transform, discrete cosine transform (DCT), and Gabor decomposition. The decomposed signal in each channel is treated according to HVS sensitivities measured in the specific channel. The error is weighted typically by a contrast sensitivity function and then adjusted by the masking effect, which reflects the reduced visibility of errors presented on the background reference signal. Finally, the errors are summarized over the image or video signal to give a single quality measure. The summation typica...