121st General Meeting of the KCS

Type Poster Presentation
Area Analytical Chemistry
Room No. Event Hall
Time 4월 19일 (목요일) 11:00~12:30
Code ANAL.P-307
Subject Analysis of Hyperspectral SORS Images using Machine Learning to Identify Substances in Containers
Authors Si Won Song, Chang Hyun Bae, Chan Ryang Park1,*, Hyung Min Kim*
Department of Bionano Chemistry, Kookmin University, Korea
1Department of Chemistry, Kookmin University, Korea
Abstract Spatially offset Raman spectroscopy (SORS) is a powerful tool that inspects the chemical information of internal components hidden in container. In the SORS measurement, the Raman signal of the internal components strongly observed as the distance between the laser irradiation point and the detection point increases. However, it is difficult to identify the Raman signal of the internal component because the Raman signal of the container material interferes with the signal of the internal component. The curve resolution methods such as multivariate curve resolution alternating least square (MCR-ALS), two-dimensional correlation analysis, self-modeling curve resolution (SMCR) are used to resolve the interference. However, the separation of the perfect spectrum is difficult because the Raman signal of the container does not disappear. In addition, in order to perform curve resolution, it is necessary to measure a lot of spectrum according to offset. In this study, we developed a device for measuring hyperspectral SORS image to confirm inner component, and classified the hyperspectral SORS images using machine learning. The hyperspectral SORS image, obtained using a large area CCD, contains a spectrum of several hundred offsets per image. We simply studied the pattern of this SORS image using machine learning to identify the constituents of the internal component. Machine learning can learn the composition of the internal component hidden in the container through learning. Therefore, our research has the potential to require accurate identification of the substances hidden in the container.
E-mail ghki007@kookmin.ac.kr