The paper presents an approach for roads detection based on synthetic aperture radar (SAR) images and road databases. The vectors provided by the database are refined using active contours (snakes). In this framework, we firstly develop a restoration filter based on the frost filter achieving an acceptable compromise between speckle elimination and lines preserving. This is followed by a line plausibility calculation step which is used to deform the snake from its initial location toward the final solution. The snake is reformulated using finite elements method. The setting of the snake parameters is not an obvious problem especially when they are tuned by trial-and-error process. We propose a new automatic computational rule for the snake parameters. Our approach is validated by a series of tests on synthetic and SAR images.

This paper presents a fully automated system for area detection based on satellite images and topographic features from a database. The surfaces issued from the database are refined using active contours (also called snakes) according to updated information provided by the multi-spectral images. The snake is reformulated using finite elements, allowing for the processing of smaller structures thus low time processing. We propose a new external energy measure through a formula by combining statistical and radiometry features. Usually the snake model is controlled by various parameters that are difficult to set by a trial-and-error process. We propose an automatic setting of these parameters and an update of the snake topology. The proposed updates allow the snake to be deformed in order to provide accurate localization. Our approach is validated in a series of tests on LANDSAT7 images.