Road Extraction from High-Resolution Remotely Sensed Image Based on Improved Ant Colony Optimization Method

Based on improved ant colony optimization method, this paper presents a new developed approach to extract road from high-resolution panchromatic remotely sensed image. The former original ant colony optimization is used to extract the road network and its detail information is not exact. This paper improved the deployment of ants and the heuristic function to extract the complete road information. With the guidance of the neighbor gray level, ants are deployed at the edge of the image and move forward for reaching the opposite side. As ants spread pheromone along their paths, roads can be extracted based on aggregated pheromone levels. This paper analyzed the practicality of this method by measuring the correctness and completeness of the extracted roads. The experimental results prove that the improved ant colony optimization method can improve the quality of the extracted roads by raising the road's correctness and retaining the road's completeness.