Feasibility of Smartphone-Based Color Matching in Fabrics for Smart Textiles Based on Color Modeling and Machine Learning

The precise matching of colors in diverse fabrics is a crucial aspect of the textile industry, traditionally achieved using costly equipment like a Spectrophotometer. However, spectropho-tometers have notable drawbacks, including high expenses, and the requirement for specialized expertise in operation. Acknowledging the potential of smartphone cameras to capture color information, this paper investigates the feasibility of utilizing smartphone images for color matching across fabrics for smart textiles. In this study, a color modeling-based model is employed to estimate color-matching attributes
$\triangle L^{*}, \triangle a^{*}$
, and
$\Delta b^{*}$
. The polarity of these attributes is compared with spectrophotometer data, as polarity is a key consideration for textile engineers. Conversely,
$CMC\triangle E^{*}$
value holds significance in determining whether to accept or reject a fabric. The machine learning-based approach in this paper achieved a training mean square error (MSE) of 0.32 and a testing MSE of 0.22 for predicting
$CMC\Delta E^{*}$
values.