AUTOMATIC FINDING OF MAIN ROADS IN AERIAL IMAGES BY USING GEOMETRIC-STOCHASTIC MODELS AND ESTIMATION

This paper presents an automated approach to finding main roads in aer ial images. The approach is to build geometric-probabilistic models fo r road image generation. We use Gibbs Distributions. Then, given an im age, roads are found by map (maximum a posteriori probability) estimat ion. The map estimation is handled by partitioning an image into windo ws, realizing the estimation in each window through the use of dynamic programming, and then, starting with the windows containing high conf idence estimates, using dynamic programming again to obtain optimal gl obal estimates of the roads present. The approach is model-based from the outset and is completely different than those appearing in the pub lished literature. It produces two boundaries for each road, or four b oundaries when a mid-road barrier is present.