Multi Kernel Support Vector Machine for Particulate Matter Estimation

Article type: Research Article

Authors: Gaur, Deepak^{; *} | Mehrotra, Deepti | Singh, Karan

Affiliations: Amity School of Engineering & Technology, Amity University, Noida, UP, India | School of Computer & Systems Sciences, Jawaharlal Nehru University, New Delhi, India

Correspondence: [*] Corresponding Author. [email protected]

Abstract: The presence of particulate matter, in the atmospheric environment, affects the health of living creatures as well as the ecosystem. Estimation of particulate matter has become one of the most challenging study for researchers. There are numerous computer techniques for the estimation of these particles. In this study, a multi kernel support vector machine (M-SVM) approach is introduced for the categorisation of particulate matter captured as digital images. Images from the archive of many outdoor scene (AMOS) have been taken for implementation purpose. The model is trained to predict the level of particulate matter captured as a digital image. An experimental model with M-SVM classification predicts the particulate matter captured as image among three levels, i.e., whether an image has a normal level, critical level or highly critical level. Simulated results were found to analyse the particulate matter with 98% of accuracy, which ensures efficient recognition of our experimental method.

Keywords: Particulate matter (PM), multi kernel support vector machine (M-SVM), image processing, archive of many outdoor scenes (AMOS) data set

DOI: 10.3233/AJW220075

Journal: Asian Journal of Water, Environment and Pollution, vol. 19, no. 5, pp. 89-95, 2022