Digital Grading the Color Fastness to Rubbing of Fabrics Based on Spectral Reconstruction and BP Neural Network

Area of Science:

• Textile Science
• Color Science
• Artificial Intelligence

Background:

• Assessing fabric color fastness to rubbing is crucial for textile quality control.
• Traditional visual grading methods are subjective and can be inconsistent.
• Digital grading offers potential for objective and automated assessment.

Purpose of the Study:

• To develop an automatic digital grading method for fabric staining color fastness to rubbing.
• To utilize spectral reconstruction technology and BP neural networks for accurate prediction.
• To compare the proposed method with existing color difference and grayscale difference methods.

Main Methods:

• Fabric samples were prepared according to ISO 105-X12 standards for rubbing.
• Visual grading was performed by experts to establish ground truth.
• Digital images were captured, and color data was extracted using spectral reconstruction.
• A BP neural network model was trained for color fastness prediction.

Main Results:

• The BP neural network model effectively predicted fabric color fastness to rubbing.
• The proposed digital grading method outperformed color difference and grayscale difference methods.
• High consistency was observed between the digital rating and visual evaluation results.

Conclusions:

• The developed digital grading method based on spectral reconstruction and BP neural networks provides a reliable and objective approach for assessing fabric color fastness to rubbing.
• This automated method offers significant advantages over traditional subjective grading techniques.
• The findings support the implementation of this technology for efficient and accurate textile quality assessment.