Land Cover Changes Analysis Based on Deep Machine Learning Technique

摘要

The methodology for solving the problem of processing of large amount of remote sensing data is proposed. The hierarchical structure of the model of deep learning method is based on neural network approach and geospatial analysis methods. This methodology was applied for high resolution land cover change mapping for Ukraine territory from 1990 to 2010. The efficiency of this approach was shown for non-arable agricultural area and changes analysis, particularly, in the eastern regions during the occupation period.