Preclassification of remote monitoring data in change detection tasks

Abstract

Detection of changes occurring on the Earth surface is one of the most important tasks of remote monitoring. It is noted that the traditional methods of detecting changes (basic components, subtraction, division, etc.) are not sufficiently effective when they are applied to the problems of remote sensing of the Earth. This is due to the fact that the images obtained by different systems under different conditions of illumination of the Earth surface have significant differences in their characteristics. To reduce the impact of qualitative differences on the result of processing, it is proposed to use a preliminary classification of objects in the images. The proposed algorithm was tested on the detection of deforestation according to the spacecraft data SPOT and Landsat Recommendations on the choice of parameters of the classification algorithm (frequency channels of images, quantitative and qualitative composition of classes, decision rules, etc.) are formulated. © 2020 American Institute of Physics Inc.. All rights reserved.