Reconstructing Spectra from RGB Images by Relative Error Least-Squares Regression

Lin, Yi-Tun and Finlayson, Graham D. (2020) Reconstructing Spectra from RGB Images by Relative Error Least-Squares Regression. Color and Imaging Conference, 2020 (28). pp. 264-269. ISSN 2166-9635

PDF (Published_Version) - Published Version
Available under License Creative Commons Attribution.

Download (9MB) | Preview


Spectral reconstruction (SR) algorithms attempt to map RGB- to hyperspectral-images. Classically, simple pixel-based regression is used to solve for this SR mapping and more recently patch-based Deep Neural Networks (DNN) are considered (with a modest performance increment). For either method, the 'training' process typically minimizes a Mean-Squared-Error (MSE) loss. Curiously, in recent research, SR algorithms are evaluated and ranked based on a relative percentage error, so-called MeanRelative-Absolute Error (MRAE), which behaves very differently from the MSE loss function. The most recent DNN approaches - perhaps unsurprisingly - directly optimize for this new MRAE error in training so as to match this new evaluation criteria. In this paper, we show how we can also reformulate pixelbased regression methods so that they too optimize a relative spectral error. Our Relative Error Least-Squares (RELS) approach minimizes an error that is similar to MRAE. Experiments demonstrate that regression models based on RELS deliver better spectral recovery, with up to a 10% increment in mean performance and a 20% improvement in worst-case performance depending on the method.

Item Type: Article
Faculty \ School: Faculty of Science > School of Computing Sciences
Related URLs:
Depositing User: LivePure Connector
Date Deposited: 19 Jun 2021 00:12
Last Modified: 30 Sep 2021 16:35
DOI: 10.2352/issn.2169-2629.2020.28.42

Actions (login required)

View Item View Item