An Electronic Equipment for Measuring Color Difference Between Tissues Based on Digital Image Processing and Neural Networks

• Autores: Calderón, José, Lipa, Benghy, Kemper, Guillermo
• Formato: artículo
• Fecha de Publicación: 2023
• Institución: Universidad Peruana de Ciencias Aplicadas
• Repositorio: UPC-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorioacademico.upc.edu.pe:10757/668095
• Enlace del recurso: http://hdl.handle.net/10757/668095
• Nivel de acceso: acceso embargado
• Materia: Color difference; Fabric color; Illuminants; Metamerism; Textile image processing; Color Difference Measurement; Fabric Selection; Illuminant Analysis; Metamerism Reduction; Enclosure Design; RGB to CIELAB Conversion; Delta E (CMC) Equation; Neural Network; Colorimeter Validation; Homogeneous Color Identification; https://purl.org/pe-repo/ocde/ford#2.02.01

This work proposes an electronic equipment aimed at measuring the color difference between fabrics for a homogeneous selection of these. The proposed method analyzes tissues with different illuminants so that, by taking measurements with a camera, the impact of effects such as metamerism can be reduced, resulting in a more accurate method that represents reality. For this, an enclosure was designed and built to allow constant lighting conditions and spotlights with standard lighting were installed. On the software side, the camera was manually configured to ensure persistent measurements over time and the colors obtained from the RGB color space were converted to CIELAB. For the calculation of the color difference, the Delta E (CMC) color difference equation was used because this is the standard used in the textile industry. Finally, a neural network was trained to estimate the color difference between the fabrics. To validate the results, the measurements obtained with the proposed equipment were compared with those obtained using a colorimeter to measure the colors of the fabrics and calculate whether their color difference is significant. The developed equipment achieved an accuracy of 95% in the correct identification of fabrics that are homogenous in color.