Applying CIECAM02 to images with unknown reference white

Abstract

The aim of this thesis was to investigate the means of applying CIECAM02 color appearance model to images with unknown reference whites. The investigation was done by estimating reference whites of digital images using four different models: white patch (WP), gray world (GW), the CIE illuminant D65 (D65) and the white point of the CRT monitor (CRT), which were tested to be used as the parameter in CIECAM02. There were three original images: "Fruits", "Wine", and "Orchids". Images reproduced via CIECAM02 with suitable reference whites would give good color matches to the original images. The result showed that there were not the reference-white estimation models that gave the best result in all cases. WP performed consistently for all images but it did not give a good result. GW gave good color matches for achromatic images such as Wine but did not work well for colorful images like Fruits. D65 provided the best result for the colorful image but did not give a good result in other cases. Overall, the white point of the testing monitor (CRT) was recommended because its performance was less dependent on images and yielded a high accuracy in all cases