Monitoring the Morphological Transformation of Beijing Old City Using Remote Sensing Texture Analysis

Abstract

This paper is concerned with the morphological analysis of Beijing old city's dynamics from 1966 to 2015. This area has been continuously submitted to internal transformations since the opening of China to a market economy. In particular, districts of small traditional houses are being replaced by large buildings, entailing a fast reorganization of the inner city. To monitor this phenomenon, we propose to characterize urban patterns with very-high-resolution images using texture analysis. To this end, dedicated urban descriptors at various scales (based on local variance, cooccurrence matrices, and wavelets) are evaluated and selected to highlight informations related to different urban patterns. These features, whose scales are essential for a reliable description, are used to highlight changes in the city of Beijing in 21 images from 1969 to 2015. The experimental results show good performances and are in accordance with expert knowledge issued from Beijing urban planning studies. About 50% of the old urban pattern has been destroyed and most of these changes occurred before 2001.