Image Quality Assessment Database for Demosaicing Artifacts
Anupama B. Lamb, Madhuri Khambete · 2018
Demosaicing is an essential process in digital color cameras for full-color image reproduction. Incorrect demosaicing process produces visible artifacts like blur, zipper artifact, moire effect, and false color. The later three artifacts are perceived as color halos. So, color halos and blur are the primary visible artifacts in the demosaiced image. In this paper, we conducted a psycho-visual experiment for assessing color halos in the demosaiced image. We decided to assess only color halos because no image quality assessment (IQA) database is available for visible color halos. Twelve undergraduate students participated in the experiment. The proposed IQA database includes 124 demosaiced images (31 reference images and 4 demosaicing algorithms). Four demosaicing techniques were selected after observing demosaiced images of 8 state-of-the-art demosaicing techniques, for visible color halos. The proposed IQA database will be utilized in developing novel IQA metrics for color halos.