Estimation of building density using Terrasar-X-Data

This study presents an approach towards the estimation of building density by utilizing the intensity data of the German space borne very high resolution SAR system TerraSAR-X. By using reference data we segmented a SAR scene of the city of Munich, Germany, on the scale of city blocks. Then we computed first order statistics and texture measures according to Haralick on the image objects. We used a random forests regression algorithm to establish the relationship between these features and the building density taken from reference data. The result is a promising coefficient of determination of 0.63.