Title :
Color Image Quality Index Based on the UIQI
Author :
Medda, Alessio ; DeBrunner, Victor
Author_Institution :
Sch. of Electr. & Comput. Eng., Oklahoma Univ., Norman, OK
Abstract :
A new full-reference color image quality index based on the universal image quality index (UIQI) of Bovik and Wang, (2002) is presented. Because the UIQI only takes into consideration distortions of the luminance component of the image, a measure on the chrominance components has been added. The chromatic image components are represented in an opponent color space that better represents the color perception of the human eye. A measure of cross-correlation between the chrominance components of the original and coded image expresses the color distortion due to image compression. Our color image quality index combines the UIQI measure and the cross-correlation measures through regression analysis using the LIVE database (H.R. Sheikh et al.). The correlation of the CIQI with the differential subjective mean opinion score (DMOS) is more than 0.8 and it is higher than the UIQI alone. A comparison between the performances of the CIQI and the PQS of Miyahara et al., (1995) is presented
Keywords :
data compression; image coding; image colour analysis; image resolution; regression analysis; LIVE database; chromatic image components; color distortion; color perception; cross-correlation measure; differential subjective mean opinion score; full-reference color image quality index; human eye; image coding; image compression; luminance component; opponent color space; regression analysis; universal image quality index; Color; Distortion measurement; Humans; Image coding; Image databases; Image quality; Image storage; Indexes; Regression analysis; Testing;
Conference_Titel :
Image Analysis and Interpretation, 2006 IEEE Southwest Symposium on
Conference_Location :
Denver, CO
Print_ISBN :
1-4244-0069-4
DOI :
10.1109/SSIAI.2006.1633753