Volcanic ash cloud extraction for RS image by combining PCA, ICA and SVM methods

摘要

A vast area of volcanic ash clouds has not only a far-reaching influence on both the global environment and climate, but also further serious threats to aviation safety. Therefore, it is of great practical significance to monitor the ash cloud condition. The extraction methods commonly applied at present cannot guarantee their consistency with the actual characteristics of remote sensing (RS) images. This paper puts forward a new volcanic ash cloud information extraction method by a combination of principal component analysis (PCA), independent component analysis (ICA) and support vector machine (SVM) algorithms, which is more oriented to remote sensing image itself. Taking the MODIS remote sensing image of Sangeang Api eruption on May 30, 2014 as an example, contrast experiments of various algorithms were conducted. The results show that the proposed method has a higher extraction precision, a better extraction effect as well as the characteristic of robustness.