Evaluation Of Fabric Color Measurement Performance And Instrument Calibration Based On LCTF Hyperspectral Imaging System

Color is an important part of color science.With the increasing development of color science,people's requirements for the accuracy of color measurement are also increasing.Especially in the field of textile,color measurement is a very significant part of textile quality control.Among the commonly used color measurement methods,the spectrophotometer method can meet the requirements of high precision,but it has some limitations in measurement region and efficiency.Aiming at these problems and considering the development direction of portable instrument in the future,a fabric color measurement scheme using hyperspectral imaging system based on Liquid Crystal Tunable Filter(LCTF)was proposed.Using this system for color measurement,more precise spatial and spectral information can be obtained,which is suitable for accurate measurement of multi-color fabrics and large area fabrics,and the measurement efficiency is higher.The main contents of this thesis are as follows:1.Evaluation of fabric color measurement performance of LCTF hyperspectral imaging system: Due to the high spatial resolution of the LCTF hyperspectral imaging system,there are gaps between the yarns in the knitted fabric images,which reduces the reflectance response of the fabric.In order to obtain more realistic reflectance of the fabric,the K-Means algorithm was used for data extraction and preprocessing which can segment yarns and gap effectively.After that,experiments were designed to study the repeatability,reproducibility and accuracy of the instrument for fabric color measurement.The experimental results show that the instrument has great repeatability and reproducibility for color measurement.However,due to the different color measurement principle and instrument structure,there exist certain errors in the color measurement results when the result of Data Color650 were taken as the standard.2.Light source consistency correction of LCTF hyperspectral imaging system:The problem of light source consistency in hyperspectral imaging systems includes the problem of uneven illumination and the difference in measurement result under different light sources.In this thesis,the standard diffusing whiteboard was photographed,and the algorithm gradient boost decision tree(GBDT)was used to fit the illumination surface to calibrate the illumination uniformity.The functional relationship between the spectral reflectance values and the spectral power distributions under different light sources was trained by deep learning neural network,and thus the spectral consistency correction model of the network for different light sources was obtained.The experimental results show that the two algorithms have good correction effect for the light sources of-halogen lamp,LED and metal halide lamp.After light calibration using the algorithms proposed,the color measurement consistency of samples can be greatly improved.3.Inter-instruments consistency correction based on improved R model: Aiming at the problem of inconsistency in the measured spectral reflectance of the same sample due to different color measurement principles and instrument structures in different instruments,an improved R model algorithm based on RBF neural network was improved.This algorithm introduced all bands as a calibration combination on the basis of the original R model,and the calibration coefficient for each band was obtained by training the RBF neural network.In the verification experiment,the model was used to correct the consistency between the two hyperspectral imaging systems and the spectrophotometer Data Color650.Experimental results show that the improved algorithm has a better effect on the spectral consistency correction of inter-instruments,and the effect is better than the original R model and the PLSR improved R model.In this thesis,the LCTF hyperspectral imaging system was used to measure the fabric color.After verifying the color measurement performance of the instrument,the consistency of the light source and inter-instruments were calibrated.The researching results of this thesis not only ensured the reliability of fabric color measurement,but also gave consideration to its consistency of measurement,which has a certain theoretical value and practical significance.