Enhancing Chromogenic Research Tools—A Cost-Effective Neural Network-Assisted Digital Camera for Spectrophotometric Analysis

A proper characterization of electrochromic films/devices needs quantifying their transmittance or reflectance modulation. Traditionally, spectrophotometers are employed for this purpose. Nevertheless, these instruments have certain limitations, such as the inability to measure large areas simultaneously. So far, various alternative experimental methods have been proposed, including multispectral cameras¹, pre-trained AI models, and techniques for obtaining reflectance spectra from images² or videos³. Despite these advances, none of these approaches can provide the same level of detail as spectrophotometers, particularly in the simultaneous characterization of entire film or device surfaces.
In this context, a neural network-based approach for reconstructing transmittance from digital images and videos has been developed⁴. A digital camera can be used as colorimeter, extracting CIELAB color coordinates from images or video files of the chromogenic materials/devices. Through a two-stage training process, the neural network can reconstruct the complete visible spectrum (400-700 nm) based on these color coordinates, achieving Mean Absolute Errors below 2.15% when compared to conventional spectrophotometers. It offers a versatile tool capable of analyzing electrochromic performance—optical contrast, switching speed, and cycling stability—across any selected part of the film or device, providing enhanced flexibility and precision for various experimental setups. In this way, this innovative method paves the way for quantitative studies on the homogeneity of electrochromic performance along big surfaces, in terms of contrast, switching speed, or cycling stability, among others.
(1) Imai, F. H.; Quan, S.; Rosen, M. R.; Berns, R. S. Digital Camera Filter Design for Colorimetric and Spectral Accuracy. In Proc. of Third International Conference on Multispectral Color Science, IS&T, Third International Conference on Multispectral Color Science; University of Joensuu, Finland, 2001; pp 13–16.
(2) Valdivieso, L. G.; Osorio, C. A.; Guerrero, J. E. Numerical Reconstruction of Spectral Reflectance Curves of Oil Painting on Canvas; Rodríguez-Vera, R., Díaz-Uribe, R., Eds.; Puebla, Mexico, 2011; p 80119P. https://doi.org/10.1117/12.903317.
(3) Agrisuelas, J.; García-Jareño, J. J.; Perianes, E.; Vicente, F. Use of RGB Digital Video Analysis to Study Electrochemical Processes Involving Color Changes. Electrochemistry Communications 2017, 78, 38–42. https://doi.org/10.1016/j.elecom.2017.04.001.
(4) Cánovas-Saura, A.; Serrano-Luján, L.; Beltrán, V.; Padilla, J. Neural Network-Based Digital Camera Spectrophotometer: Application to Chromogenic Technologies Characterization. ACS Appl. Opt. Mater. 2024, acsaom.4c00180. https://doi.org/10.1021/acsaom.4c00180.