Fabric Tactile Prediction Method Based on Spider Diagram

Ruifeng Xie, Shuyang Ding, Zeyu Cheng, Lizhuang Ma, Yanzhu YangSensors 2025 · DOI

TL;DR

This paper introduces a novel method for predicting fabric tactile sensations using a "spider diagram" approach. By measuring five key physical properties of fabrics and visualizing them on a radar chart, the researchers developed a model that accurately predicts subjective tactile experiences, aiming to improve online fabric shopping.

What Problem Does This Paper Address?

The textile industry, particularly online retail, faces a significant challenge in conveying the tactile feel of fabrics to consumers. While visual cues and descriptions are common, they often fall short of capturing the nuanced tactile experience that is crucial for comfort and purchasing decisions. Previous research has focused on measuring individual fabric physical properties but has often neglected to establish clear correlations between these objective measurements and subjective human perception. This gap makes it difficult for consumers to assess fabric feel online, leading to potential dissatisfaction. This study aims to bridge this gap by developing a predictive model that links objective fabric properties to subjective tactile sensations.

Methods

The researchers employed a multi-faceted approach:

  • Subjective Evaluation: 50 volunteers participated in a controlled experiment to subjectively rate fabric samples on cool-warmth, smoothness, and softness. This provided a baseline of human perception.
  • Objective Physical Property Measurement: Five key physical properties were measured for 15 fabric samples:
  • Roughness (Kv): Measured using a POD tribometer, analyzing vibrational signals during friction.
  • Fullness (Kw): Also derived from the POD tribometer's vibrational analysis.
  • Bendability (B): Measured using the KES-FB2-A pure bending tester.
  • Elasticity (K): Determined using a custom tensile testing device.
  • Thermal Conductivity (ΔK): Assessed using a self-made device measuring temperature differences.
  • Spider Diagram Construction: The normalized values of these five physical properties were plotted on a radar chart (spider diagram) for each fabric. This visual representation allows for a holistic view of a fabric's tactile profile.
  • Predictive Modeling: A mathematical model was developed to correlate the spider diagram's area and the interaction of physical properties with the subjective evaluation scores. This involved a least-squares nonlinear fitting approach to determine the influence of each property and their interactions.

Datasets

The study utilized a dataset of 15 commercially available fabric samples. These samples were categorized based on:

  • Yarn Weight: Ranging from 3.5 to 15 oz/yd².
  • Weave Pattern: Including canvas, gauze, satin, and twill.
  • Material: Comprising natural materials (cotton, silk, wool) and composites (silk/cotton, silk/elastane).

This curated selection aimed to provide a diverse yet representative sample for analysis. No pre-existing benchmark datasets were explicitly mentioned as being used for training or evaluation, suggesting this study introduces its own experimental dataset.

Key Insights and Ideas

  • The Power of Visualization: The spider diagram proves to be an effective tool for visualizing complex fabric tactile properties in a single, intuitive graphic. This makes it easier to compare and understand the differences between fabrics.
  • Interactions Matter: The study highlights that the interaction between physical properties is crucial for predicting tactile sensation, not just the individual properties themselves. This moves beyond simple additive models.
  • Key Predictors Identified: Bendability (B), surface roughness (-Kv), and fullness (Kw) were found to be the most significant factors influencing the overall tactile sensation, as indicated by higher R-squared values in the predictive model.
  • Subjective vs. Objective Alignment: The developed model demonstrates a strong correlation between objective physical measurements and subjective human perception, validating the approach.
  • Bridging the Online Shopping Gap: The spider diagram and predictive model offer a concrete, data-driven way to communicate fabric feel online, moving beyond vague descriptors.

Key Results

  • The developed predictive model achieved a high accuracy, with an adjusted R-squared value of 0.87332, indicating that the model explains a significant portion of the variation in subjective tactile ratings.
  • The model identified bendability, surface roughness, and fullness as the primary drivers of subjective tactile perception.
  • The spider diagram analysis revealed that fabrics with larger spider diagram areas generally received higher subjective evaluations, suggesting a correlation between overall physical property profile and perceived comfort.
  • A comparison of predicted versus subjective evaluation scores showed a high degree of fit, validating the reliability of the proposed model.

Limitations and Future Work

  • Limited Sample Size: The study used 15 fabric samples. While diverse, a larger and more varied dataset could further enhance the model's generalizability.
  • Subjectivity of Human Judgment: While efforts were made to mitigate bias (e.g., noise-canceling headphones, large volunteer group), subjective evaluations inherently carry some degree of variability.
  • Future Work: The authors suggest that this model can be a scientific foundation for textile design and development, enabling manufacturers to optimize product design based on consumer preferences. Further research could explore expanding the number of physical properties considered and refining the predictive model for even greater accuracy and broader applicability. The model's ability to facilitate more effective communication of tactile information in online shopping environments is also a key area for future development.

Summary updated

Abstract

The detection and quantification of fabric tactile sensations are crucial in textile production and marketing as they are closely linked to textile comfort and serve as key criteria for consumers when selecting fabrics. Previous studies have predominantly focused on measuring the physical properties of fabrics, often neglecting correlations between these parameters and tactile sensations. This oversight complicates customers' ability to assess the tactile experience of fabrics during online purchasing. This study first obtained subjective evaluations of three types of fabric tactile sensations through experiments involving volunteer participants. Subsequently, five objective physical properties that characterize fabric tactile properties were proposed and experimentally tested on 15 fabric samples categorized by yarn weight, weave pattern, and material. A fabric tactile spider diagram was created by normalizing the values of the five physical properties across the 15 fabric samples. The grading of the physical properties was then performed based on the proposed evaluation index. These spider diagrams were compared with the subjective evaluation results to analyze the physical properties that most significantly influenced subjective perception, ultimately leading to the development of a highly reliable fabric touch prediction model.
BibTeX
@inproceedings{xie2025fabric,
  author = {Ruifeng Xie and Shuyang Ding and Zeyu Cheng and Lizhuang Ma and Yanzhu Yang},
  title = {{Fabric Tactile Prediction Method Based on Spider Diagram}},
  booktitle = {Sensors},
  year = {2025},
  doi = {10.3390/s25103187},
}
Figure