A measure for image quality
Le Phu Dung, Srinivasan Bala, Mohammed Salahadin, Santosh Kulkarni, Wilson Campbell · 1998
This paper describes a new measure for calculating the error that has been introduced when an image is modified which can be used to compare the quality of images.It is based on the following principle:The quality of an image should be measured locally and globally.This implies that:(1) Since the change of each pixel value affects the quality of the image, and the overall quality is dependent on the total number of pixels that were changed, the total number of pixels that were changed should hence be used in evaluating the quality of the image.(2) The change of each pixel value first affects the quality of a small part of the image that directly contains the pixel, and the change of that part of the image affects the quality of the whole image.This suggests that image quality should be evaluated part by part rather than pixel by pixel.Experimental results show that this proposed measure performs better than the signal-to-noise ratio and colour-based methods.It not only works well when the other two methods work but also works where the other two methods fail.mission to make digital/hard copy of all or part of this work for personal or :~room use is granted without fee provided that copies are not made or ributed lbr profit or commercial advantage, the copyright notice, the title of the ,lication and its date appear, and notice is given that copying is by permission of M, Inc, To copy otherwise, to republish, to post on servers or to redistribute to ;, requires prior specific permission and/or a the.