An approximated Snake Function for Road Extraction from digital images

Abstract

This paper proposes an optimized mathematical model (Snake-ant) for linear feature extraction from satellite images. The model first uses the Ant Colony Optimization (ACO) to establish a pheromone matrix that represents the pheromone information at each pixel position of the image, according to the movements of a number of ants which are sent to move on the image. Next pheromone matrix is used in the snake model as external energy to extract the linear features like roads edges in image. Snake is a parametric curve which is allowed to deform from some arbitrary initial location toward the desired final location by minimizing an energy function based on the internal and external energy. Our approach is validated by a series of tests on satellite images.