Back to results list
Please use this identifier to cite or link to this item:
|Title:||An investigation of texture effects on instrumental and visual colour difference evaluation||Authors:||Shao, Sijie||Keywords:||Hong Kong Polytechnic University -- Dissertations
|Issue Date:||2006||Publisher:||The Hong Kong Polytechnic University||Abstract:||The need for accurate colour measurement has been ever increasing for the establishment of a reliable colour quality control system in a number of industries, especially in the textile and garment industry. Colour measurement is achieved using instruments such as colorimeter, spectrophotometers. These instruments are useful for measuring large homogeneous samples. In this thesis, the first goal of this work is to build a colorimetric and multispectral imaging acquisition system to measure CIE trichromatic coordinates or recover the reflectance information at each pixel location of texture images. In addition, the reliability of the colorimetric imaging system and spectrophotometers used for colour difference was studied in comparison with the visual assessments of physical texture samples under the controlled viewing conditions. Although, there was a limit for the absolute measurement of texture colour patches using colorimetric imaging system, the results showed that the colorimetric imaging system was satisfactory for the relative measurement of texture colour sample for determination of colour difference. For the multispectral imaging system, an independent component analysis (ICA) method was first used to estimate the spectral basis functions. The comparison of this method with the principle component analysis (PCA) method was made. The results showed that five spectral basis functions i.e., principle components or independent components estimated from PCA and ICA respectively, are sufficient to recover the spectral information of texture samples. Several estimation methods of the spectral sensitivity of the five-band multispectral imaging system were compared. The quadratic optimization method with the nonnegative and regularization method provided a higher accuracy of the estimation of spectral sensitivity in comparison with the general pseudo-inverse (PI) and principle eigenvectors (PE) methods. Wiener estimation method was used to recover the spectral information from the five-band multispectral camera responses. The results show that the estimated spectral information from fine texture images by the multispectral imaging system is in agreement to the measurements by the spectrophotometer. In instrumental colour control process, the results of colour difference calculations between standards and their matches often determine whether these matches are passes or fails. There is no doubt that the pass/fail decisions significantly affect the colour quality of products and the production lead-times as well as the production costs. For accurate quantification of colour difference, a reliable colour-difference formula is required. An ideal colour difference formula should accommodate the changes of viewing parameters. The effects of sample size, background colours, luminance, lightness of sample, and the magnitude of colour differences have been carefully investigated previously on the colour difference evaluation. Surface texture is another parametric factor significantly influencing colour difference evaluation and thus is very important for industrial colour control. Presently, no quantitative relationship between the texture parameter and its effect on the visual colour difference has been established. Therefore, no correction factor or guideline has been suggested by the CIE. This leads to the second goal of this work, which is the study of the texture effect on small, medium and large visual colour difference evaluation. The differently woven or knitted cotton fabric samples and synthesized texture samples generated by texture mapping techniques at five specified CIE colour centers were used in the study. Various phases of psychophysical experiments were conducted using grey scale methods. The accumulated data was firstly used to quantify the relationship between the visual colour difference and the parametric texture effects. Two texture parametric effects, named by texture strength contrast and the product of texture strength product, were extracted from the histogram distribution of 2-D Gabor Wavelet Transform of two colour texture images and were used to investigate their relationships to the visual colour difference of texture colours. Secondly, a new texture colour difference model (TCDM) was built to predict the visual colour difference of two texture colours with a different texture structures. The visual results were used to test the performance of the TCDM models as well as five other colour difference formulae, namely, CIELAB, CMC, BFD, CIE94 and CIEDE2000 with the built-in optimized parametric factors KL, Kc, and KH. The test results clearly indicate that the TCDM model can predict the visual colour difference of the texture samples and it is superior to those five colour difference formulae with the optimized parametric factors. The results also pointed out that the currently widely used method of instrumental colour measurement and the colour difference formulae are unable to predict the visual colour difference assessment of texture colours.||Description:||xxv, 331 leaves : col. ill. ; 30 cm.
PolyU Library Call No.: [THS] LG51 .H577P ITC 2006 Shao
|URI:||http://hdl.handle.net/10397/2278||Rights:||All rights reserved.|
|Appears in Collections:||Thesis|
Show full item record
Files in This Item:
|E-thesis_Link.htm||For PolyU Users||162 B||HTML||View/Open|
|b20592760.pdf||For All Users (Non-printable)||5.83 MB||Adobe PDF||View/Open|
Citations as of Sep 16, 2018
Citations as of Sep 16, 2018
Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.