Abstract—Most Image Quality Assessment (IQA) methods measure the overall image quality including all visible and non-visible errors. They often do not correlate well with visual assessment particularly when non-visible errors are large in proportion. In particular, color artifacts are a crucial factor in visual assessment, but they might only have a small contribution to the total errors as they are often minorities. Hence, it is desirable to specifically measure color artifacts alone, excluding other errors. One application for such measurement is that color artifacts are the main visible errors produced by Color Filter Array (CFA) demosaicking algorithms. By formalizing the perception that color artifacts manifest as distinct visual color variation from their original and surrounding colors, a novel IQA method, namely Normalized Color Variation (NCV), is proposed specifically for locating and quantifying color artifacts. It gives a NCV index which is a measure of the degree of color artifacts. It has been shown that our proposed NCV method based on the formalization of the perception of color artifacts correlated well with our visual perception. Its NCV index has proven to be a good indicator of the degree of color artifacts and is virtually independent of other errors.
Index Terms—color artifacts, Normalized Color Variation (NCV), image quality assessment, hue assumption, CFA demosaicking
Cite: Omar Shakar, Jim S. Jimmy Li, and Sharmil Randhawa, "Quantification of Color Artifacts for the Evaluation of Color Filter Array Demosaicking," Journal of Image and Graphics, Vol. 6, No. 1, pp. 48-53, June 2018. doi: 10.18178/joig.6.1.48-53
Copyright © 2012-2020 Journal of Image and Graphics, All Rights Reserved