Following the first promising preliminary results, we study different strategies for the extension to color images of the Wavelet Normalised Root Mean Squared Error (WNRMSE), a new image similarity assessment metric, previously introduced and extensively studied by the authors for monochromatic images. The theoretical properties of the resulting color image metrics are studied, and the different extension strategies are tested and compared on the standard benchmark database TID2013.

This presentation is part of Contributed Presentation “CP5 - Contributed session 5”

Authors:

Silvia Bertoluzza (CNR, IMATI "Enrico Magenes")

Riccardo Amadeo (Department of Electrical, Computer and Biomedical Engineering, University of Pavia)

Maria Grazia Albanesi (Department of Electrical, Computer and Biomedical Engineering, University of Pavia)

Keywords:

color image, computer vision, similarity assessment