Nonwovens structure measurement based on NSST multi-focus image fusion

Area of Science:

• Materials Science
• Image Processing
• Textile Engineering

Background:

• Digital optical microscopy struggles with nonwoven fabric thickness, limiting clear visualization of all fibers in a single image due to shallow depth of field.
• Accurate structural analysis of nonwovens requires clear imaging of individual fibers and pores.

Purpose of the Study:

• To develop and validate a multi-focus image fusion algorithm for nonwoven fabrics.
• To improve the clarity and completeness of fiber visualization in a single image.
• To enable automated measurement of nonwoven fabric structural parameters.

Main Methods:

• A novel multi-focus image fusion algorithm based on non-subsampled shearlet transform (NSST) was developed.
• High-frequency sub-bands were fused using a large absolute value rule, and low-frequency sub-bands using a large regional variance rule.
• Hough transform and image preprocessing were applied to the fused image for automated diameter and orientation measurements; pore identification was used for porosity measurement.

Main Results:

• The proposed NSST-based fusion algorithm significantly improved fused image quality compared to other methods, as indicated by image quality evaluation metrics.
• All fibers within the nonwoven fabric were clearly visualized in a single fused image.
• Automated and accurate measurements of fiber diameter, orientation, and porosity were achieved.

Conclusions:

• The developed multi-focus image fusion technique effectively addresses the depth-of-field limitations in digital optical microscopy of nonwovens.
• The algorithm enables rapid and convenient automated measurement of key structural parameters in nonwoven fabrics.
• This image processing approach offers a valuable tool for the characterization and analysis of nonwoven materials.