Entity based image analysis for impervious and pervious surface mapping using remote sensing imagery

Abstract

Rapid urbanization and residential development leads to increase in amount of Impervious surface area, which badly affects the natural environment. Impervious surfaces are artificial or manmade surfaces which are highly resistant to infiltration of water. Estimating pervious and impervious region can be useful to assessing the environmental quality and the outcome of it can be used to manage urbanization and its impact on environment. The pervious and impervious surface estimation can be done based on per pixel, sub pixel, parcel based and super pixel based techniques. Urban impervious surface detection using remote sensing with high spatial resolution imagery is difficult due to spectral, temporal, and spatial variations of urban areas. Spectral variations are the presence of shadows and the presence of unknown or mixed pixels which are considered as statistical noise. Automated mapping of impervious surfaces with reasonable accuracy in an urban environment is one of the most difficult issues which has still not been addressed completely using remote sensing. This study investigates possible avenues for automation and assesses newlinethe current bottlenecks during processing of imagery using Entity-Based Imagery Analysis (EBIA). EBIA incorporates spectral and spatial information in its analysis. EBIA also operates at a better defined level of image objects, which consist of multiple pixels grouped according to a set of pre-defined criteria rather than single pixels used in pixel-based classification. Impervious surface mapping can be done with many conventional approaches newline newline