In the automatic assessment of image quality we obtained a high accuracy in
the classification of image degradations in a manner that is widely indepe
ndent of scene content. Using an all-digital ring-wedge detector system com
bined with neural-network software, we conducted several experiments in whi
ch the end goal is to classify images according to numerical quality scales
. Experiments are presented to stress the importance of both local and glob
al image quality assessment. Two databases of degraded Images were prepared
. One uses five levels of Gaussian blur to simulate depth of field. The oth
er was prepared with lossy compression and recovery with artifacts generate
d by a JPEG (Joint Photographic Experts Group) compression algorithm. In qu
antitative terms our best sorting of Gaussian blur without knowledge of the
original scene was to an accuracy of 96%. For degradation using JPEG we ob
tained an accuracy of 95% without knowledge of the original and 98% when th
e original scene is available as a reference. (C) 2000 Optical Society of A
merica OCIS codes: 070.5010, 070.2590, 100.2000, 200.4260, 110.3000.