Abstract

We evaluate three link functions (square root, logit, and copula) and Matérn kernel in the kernel-based estimation of reflectance spectra of the Munsell Matte collection in the 400–700 nm region. We estimate reflectance spectra from RGB camera responses in case of real and simulated responses and show that a combination of link function and a kernel regression model with a Matérn kernel decreases spectral errors when compared to a Gaussian mixture model or kernel regression with the Gaussian kernel. Matérn kernel produces performance similar to the thin plate spline model, but does not require a parametric polynomial part in the model.

© 2013 Optical Society of America

Spectral imaging using consumer-level devices and kernel-based regression

Ville Heikkinen, Clara Cámara, Tapani Hirvonen, and Niko Penttinen
J. Opt. Soc. Am. A 33(6) 1095-1110 (2016)

Evaluation and unification of some methods for estimating reflectance spectra from RGB images

Ville Heikkinen, Reiner Lenz, Tuija Jetsu, Jussi Parkkinen, Markku Hauta-Kasari, and Timo Jääskeläinen
J. Opt. Soc. Am. A 25(10) 2444-2458 (2008)

Regularized learning framework in the estimation of reflectance spectra from camera responses

Ville Heikkinen, Tuija Jetsu, Jussi Parkkinen, Markku Hauta-Kasari, Timo Jaaskelainen, and Seong Deok Lee
J. Opt. Soc. Am. A 24(9) 2673-2683 (2007)

Evaluating logarithmic kernel for spectral reflectance estimation—effects on model parametrization, training set size, and number of sensor spectral channels

Timo Eckhard, Eva M. Valero, Javier Hernández-Andrés, and Ville Heikkinen
J. Opt. Soc. Am. A 31(3) 541-549 (2014)

Spectral reflectance estimation from camera responses by support vector regression and a composite model

Wei-Feng Zhang and Dao-Qing Dai
J. Opt. Soc. Am. A 25(9) 2286-2296 (2008)