Image processing and the problem of quantifying image quality
Huib de Ridder · 2002
The problem of quantifying perceived quality of processed images is traditionally solved by associating quality with the deviation from the undistorted original. This paper presents experiments challenging two underlying assumptions, namely (1) the original always possesses the highest quality and (2) quality cannot be determined without a direct comparison between the original and processed image. In doing so, percentage scaling is introduced as a scaling procedure for evaluating multiply impaired images.