Optical coherence tomography image dataset of textile fabrics

Published: 22 October 2021| Version 1 | DOI: 10.17632/kddwp4k7ff.1
Contributors:
Metin Sabuncu,
Hakan Ozdemir

Description

Optical coherence tomography (OCT) images of textile fabrics were taken. The OCT scans of fabrics that consisted of solely wool, cotton or polyester were recorded. For each material type three different fabrics were measured with an OCT imaging system. In total 9 different fabrics (3 wool, 3 polyester and 3 cotton) were each scanned at least 120 times on separate places on their surface. The zip file contains all the images in separate folders. The folder Polyester2 contains OCT images of the second polyester fabric sample. To have roughly uniform data across samples, the scans were fixed at 2mm for each image and recorded in portable network graphics format. The fabric data were separated into three categories. Group 1, 2 and 3 consisted of fabrics made out of only cotton, wool and polyester, respectively. These can be utilized as the labeled dataset for the classes in the deep learning training. We make this labelled OCT fabric type data set public. Moreover this public data can be used to test existing machine learning algorithms. Automated material classification and recycling can be accomplished via these deep learning networks. The uploaded data consists of raw OCT images, all in portable network graphics format in 9 separate folders. Each folder contains OCT scans taken from random places over one of the fabrics only. Therefore folders Wool1, Wool2 and Wool3 each consist of OCT scans of three separate fabrics that consisted of purely wool fibres. Similarly folders Cotton1, Cotton2 and Cotton3 consist of OCT scans of three separate fabrics that consisted of purely cotton fibres. Again folders Polyester1, Polyester2 and Polyester3 consist of OCT scans of three separate fabrics that consisted of purely polyester. The scans are raw OCT-B image files saved through the ThorImage software program of the Thorlabs OCT Imaging System. Note that no extra image processing or filtering was performed on the raw OCT images. Each OCT scan is a separate image, with the time and date of the scan stamped in the corresponding portable network graphics file name. The file in Folder Wool1 with name OCTImage 053324_07172020.png therefore corresponds to an OCT scan of the first wool fabric taken at 05:33:24 on July 17, 2020. Wool2 folder consists of OCT scans of only the second fabric woven from wool. Correspondingly; folders Cotton1, Cotton2 and Cotton3 consist of OCT scans of three separate fabrics that consisted of purely cotton fibres. Likewise; folders Polyester1, Polyester2 and Polyester3 consist of OCT scans of three separate fabrics that consisted of purely polyester.

Files

Steps to reproduce

The OCT images were taken by a Thorlabs CAL110C1 imaging system, that employed a laser diode with central wavelength at 930 nm [1]. The diode generates broad band photons, that allows for speckle free imaging [2,3,4]. These scans are then captured by employing the ThorImage OCT software [5]. The fabric samples to be scanned are put in the sample arm. The 2 Dimensional image scans known as OCT-B scans are obtained by scanning the light beam over the surface of the sample and adding corresponding OCT-A images sequentially [6]. The scan length was fixed to 2 mm for all OCT-B images taken from the 9 different fabric samples [7]. This allowed to have roughly uniform data across samples. Since most deep learning algorithms require at least 100 images per class for the training, we made sure to take between 120-200 OCT scans per fabric. Each sample was put into the sample arm and measured by the OCT modality individually and recorded in a separate folder in portable network graphics format [8]. The OCT images are raw OCT-B files saved through the ThorImage software package [9], and no extra image processing or filtering was performed on the data [10]. References: [1] M. Sabuncu and H. Ozdemir, Classification Of Material Type From Optical Coherence Tomography Images Using Deep Learning, International Journal of Optics, 2021. [2] M. Sabuncu, M. Akdoğan, Photonic imaging with optical coherence tomography for quality monitoring in the poultry industry: A preliminary study, Rev. Bras. Cienc. Avic. 17 (2015) 319–324. [3] I. Yılmazlar and M. Sabuncu, Speckle noise reduction based on induced mode Hopping in a semiconductor laser diode by drive current modulation, Opt. Laser Technol., vol. 73, pp. 19–22, 2015. [4] I. Yilmazlar and M. Sabuncu, Implementation of a current drive modulator for effective speckle suppression in a laser projection system, IEEE Photonics J., vol. 7, no. 5, pp. 1–6, 2015. [5] M. Sabuncu, H. Özdemir, Recognition of fabric weave patterns using optical coherence tomography, J. Text. Inst. 107 (2016) 1406–1411. [6] M. Sabuncu, H. Ozdemir, Contactless measurement of fabric thickness using optical coherence tomography, The Journal of the Textile Institute. (2021) 1901457. [7] M. Sabuncu, H. Ozdemir, M. U. Akdogan, Automatic identification of weave patterns of checked and colored fabrics using optical coherence tomography, IEEE Photonics J. 9 (2017) 1–8. [8] A.G. Turk, M. Sabuncu, M. Ulusoy, Evaluation of adaptation of ceramic inlays using optical coherence tomography and replica technique, Braz. Oral Research, vol. 32, pp. 1-10, 2018. [9] M. Sabuncu and H. Özdemir, Recognition of weave patterns of striped fabrics using optical coherence tomography, Fibres Text. East. Eur., vol. 26, no. 3(129), pp. 98–103, 2018. [10] M. Sabuncu, M. Akdoğan, Utilizing optical coherence tomography in the nondestructive and noncontact measurement of egg shell thickness, Scientific World Journal. (2014) 205191.

Institutions

Dokuz Eylul Universitesi

Categories

Image Processing, Wool, Machine Learning, Optical Coherence Tomography, Polyester, Textile Fiber, Cotton, Recycling, Textile Engineering, Deep Learning, Textile Technology

License