An investigation of colour measurement of yarn dyed fabrics based on the multispectral imaging system

Pao Yue-kong Library Electronic Theses Database

An investigation of colour measurement of yarn dyed fabrics based on the multispectral imaging system

 

Author: Luo, Lin
Title: An investigation of colour measurement of yarn dyed fabrics based on the multispectral imaging system
Degree: Ph.D.
Year: 2015
Subject: Spectroscopic imaging.
Textile fabrics -- Coloring.
Hong Kong Polytechnic University -- Dissertations
Department: Institute of Textiles and Clothing
Pages: xxv, 211 leaves : color illustrations ; 30 cm
Language: English
InnoPac Record: http://library.polyu.edu.hk/record=b2806852
URI: http://theses.lib.polyu.edu.hk/handle/200/7927
Abstract: Commercial spectrophotometers and current state-of-the-art digital camera imaging systems are unable to measure the spectral reflectance of yarn dyed fabrics, the former can only measure the average reflectance of an area and the latter can only measure the tristimulus values derived from the camera’s RGB responses. Multispectral imaging systems, on the other hand, have the potential to measure the reflectance of a multi-colour object, such as yarn dyed fabrics, since they can record both of the spectral and spatial information of a sample. In this thesis, colour measurement of yarn dyed fabrics based on the multispectral imaging technique is studied. The major factor restraining multispectral imaging systems from application in textile is the difficulty of correlation the measurement results of a yarn dyed fabric to the true colours of the yarns. The spectral response of multispectral imaging systems to yarn dyed fabrics is dramatically affected by irregular 3D shapes of yarns, inter-reflection between neighbouring yarns, and interstices between weft and warp yarns. In this thesis, a novel reflection model is first proposed to estimate the interaction between light and a yarn dyed fabric surface. Surface texture, illumination occlusion and inter-reflection are taken into account. The reflection model is then verified by reducing the influence of texture on spectrophotometric colour. Derived from the proposed reflection model, reflectance and tristimulus values of yarn dyed fabrics with different texture structures are linear. The linear relationship in the reflectance space can be used to estimate a theoretical reflectance which discounts the influence of texture. Experimental results show that the impact of texture on colour for yarn dyed fabric samples in four colour centres and twenty-one texture structures can be reduced by 79%, 55%, 71% and 57%, respectively. Based on the proposed reflection model, multispectral imaging colour measurement of yarn dyed fabrics are achieved through a series of image processing techniques, namely, colour region segmentation, solid-colour and multi-colour region detection, and weft and warp yarn segmentation. Firstly, a yarn dyed fabric image is partitioned into dominant colour regions by a Gaussian model. The Gaussian model is used to reconstruct the CIELAB colour histograms of dominant colour regions from those of yarns. A hierarchical segmentation structure is then devised to obtain dominant colour regions by combining histogram segmentation results in three colour channels. Experimental results shows the proposed approach has excellent performance for dominant colour region segmentation with high computational efficiency. Secondly, a dominant colour region is detected as solid-colour or multi-colour by CIExyY histogram distributions. Derived from the proposed reflection model, the CIExyY histogram of a multi-colour yarn dyed fabric accords with a combination of two Gaussian distributions, whereas for that of a solid-colour yarn dyed fabric, it correlated to one Gaussian distribution. Experiments on real yarn dyed fabric samples demonstrate that solid-colour and multi-colour yarn dyed fabric regions can be distinguished in terms of CIExyY histogram distribution. Finally, a multi-colour yarn dyed fabric is segmented to weft and warp yarns by a modified K-means clustering method. Experimental results indicate that the proposed method can segment weft and warp yarns in yarn dyed fabric images, with both high segmentation accuracy and fast processing speed.
In addition, the proposed reflection model can be utilized to accomplish multispectral imaging colour measurement of single yarns. Single yarns are the elemental weaving components of yarn dyed fabrics and have much simpler structures. The multispectral imaging colour of a single yarn is not affected by surface texture and inter-reflection. The colour measurement of single yarns is achieved by two steps. Firstly, a single yarn is segmented from backgrounds by an image difference method. Secondly, the reflectance of a single yarn can be specified by different weighting methods. Experimental results show that multispectral imaging colour measurement of single yarns can achieve a repeatability of 0.1185 CMC(2:1) units and a spatial reproducibility of 0.2827 CMC(2:1) units. Experimental results also show that single yarns measured by multispectral imaging systems can accomplish the similar colour matching results as solid-colour yarn dyed fabrics measured by spectrophotometers. Finally, an optical-based approach is proposed to explore the relation between the multispectral imaging colour of a single yarn and the spectrophotometric colour of the corresponding yarn card. A colour mapping equation between the single yarn and corresponding yarn card can be then found by the simplex optimal method. Experimental results show that colour difference between single yarns and corresponding yarn cards reduces from 2.97 to 1.20 CMC(2:1) units for 50 pairs of training samples and from 3.09 to 1.37 CMC(2:1) units for 50 pairs of testing samples.

Files in this item

Files Size Format
b2806852x.pdf 6.416Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information