A novel approach to thermochromic liquid crystal calibration using neural networks

This source preferred by Jon Cobb and Venky Dubey

Authors: Grewal, G.S., Bharara, M., Cobb, J.E., Dubey, V.N. and Claremont, D.J.

http://www.iop.org/EJ/abstract/0957-0233/17/7/033/

Journal: Measurement Science and Technology

Volume: 17

Pages: 1918-1924

ISSN: 0957-0233

DOI: 10.1088/0957-0233/17/7/033

Liquid crystal thermography (LCT) is a common surface temperature measurement technique. Typically, the colour response is calibrated against temperature by building an analytical relation between the temperature and the hue of the colour. A suitable polynomial fit is then used to describe this relation after removing the discontinuity in the hue. The variability of hue at each calibration point determines the temperature resolution. However, this technique does not take into consideration the variability in R, G and B intensities used to determine the hue, leading to uncertainty in the measured temperature. This paper describes a novel technique using neural networks to calibrate thermochromic liquid crystal (TLC) material and compensate for high variability in RGB intensities along with other sources of noise in the data. A TLC-based temperature measurement system and calibration results are presented. In our measurements, the lighting intensity (8-bit mean intensity of black surface ± standard deviation) is changed from a minimum of 16.65 ± 2.30 to a maximum of 31.41 ± 3.85. The neural networks were trained on the steady-state TLC calibration system. The results indicate that the neural networks can cope with the variation in lighting by merging the shifted hue curves into a single curve determined by the regression analysis of the test data. Performance characteristics studied on various network configurations relevant to the analysis are described. This approach may be useful in developing liquid crystal thermography for various biomedical applications.

This data was imported from Scopus:

Authors: Grewal, G.S., Bharara, M., Cobb, J.E., Dubey, V.N. and Claremont, D.J.

Journal: Measurement Science and Technology

Volume: 17

Issue: 7

Pages: 1918-1924

eISSN: 1361-6501

ISSN: 0957-0233

DOI: 10.1088/0957-0233/17/7/033

Liquid crystal thermography (LCT) is a common surface temperature measurement technique. Typically, the colour response is calibrated against temperature by building an analytical relation between the temperature and the hue of the colour. A suitable polynomial fit is then used to describe this relation after removing the discontinuity in the hue. The variability of hue at each calibration point determines the temperature resolution. However, this technique does not take into consideration the variability in R, G and B intensities used to determine the hue, leading to uncertainty in the measured temperature. This paper describes a novel technique using neural networks to calibrate thermochromic liquid crystal (TLC) material and compensate for high variability in RGB intensities along with other sources of noise in the data. A TLC-based temperature measurement system and calibration results are presented. In our measurements, the lighting intensity (8-bit mean intensity of black surface standard deviation) is changed from a minimum of 16.65 2.30 to a maximum of 31.41 3.85. The neural networks were trained on the steady-state TLC calibration system. The results indicate that the neural networks can cope with the variation in lighting by merging the shifted hue curves into a single curve determined by the regression analysis of the test data. Performance characteristics studied on various network configurations relevant to the analysis are described. This approach may be useful in developing liquid crystal thermography for various biomedical applications. © 2006 IOP Publishing Ltd.

This data was imported from Web of Science (Lite):

Authors: Grewal, G.S., Bharara, M., Cobb, J.E., Dubey, V.N. and Claremont, D.J.

Journal: MEASUREMENT SCIENCE AND TECHNOLOGY

Volume: 17

Issue: 7

Pages: 1918-1924

eISSN: 1361-6501

ISSN: 0957-0233

DOI: 10.1088/0957-0233/17/7/033

The data on this page was last updated at 04:42 on November 17, 2017.