Particle Swarm Optimization for Road Extraction inSAR Images

摘要

The paper proposes a new method for road extraction in SARrnimages. We regard that the road in SAR images can be represented by the Bsplinerncurve. Firstly, we manually select the road’s extremities. Secondly, werncalculate the each pixels’s road membership value using local road detector inrnthe original SAR images. Thirdly, with particle swarm optimization that is onernof the most powerful methods for optimization problem we obtain the optimalrnB-spline control points from the result of road detection. Finally, according tornthe optimal B-spline control points, we obtain the B-spline curve that is thernresult of road extraction. The experimental result shows that the method in thernpaper can accurately extract the road.