Use of directional variance for urban area analysis on simulated Spot 5 images

CNES is due to launch the SPOT 5 satellite during 2001. One of its particularities, compared to the previous generations of SPOT satellites, will be its high resolution in panchromatic mode (5m and Supermode 2.5m). Before launch, CNES has to validate the choices made for the sensors with the use of simulated images. The authors's study has been made as part of the use of SPOT 5 panchromatic images for urban area analysis. The aim of the study is to extract urban areas and roads from panchromatic simulated SPOT 5 images. The proposed method combines a textural feature extraction with segmentation by edge detection in order to improve the final classification. At first, they define a textural feature based on the directional variance of the image. Then they use classification algorithms on this feature, combined with an edge detection operator, to extract the urban areas and the roads from the image.