As manual extraction of road is time consuming and it relies on operator’s efficiency, automatic extraction of road from remotely sensed data has become the subject of extensive research for the past decade. In this paper the task of automatically extracting road remotely sensed image is explained. Road can be extracted from colour image by determining its spectral signature. As there are more than 16 million colours available in any given colour image analysing the image based on its colour components is a complex task. Even if the road spectral signature is extracted, it is difficult detect road alone since soils and roads have the same spectral signature. A hybrid technique has been developed to extract road. First the image is preprocessed to remove the noise. Then the road spectral signature is extracted using Artificial Neural network (ANN). Contour tracing algorithm and mathematical morphology are used to remove the non road segments.

Keywords: Road extraction, ANN, LoG Filter, Contour tracing, Granulometry, Mathematical Morphology