Vehicle Detection and Orientation Estimation Using the Radon Transform

Determining the location and orientation of vehicles in satellite and airborne imagery is a challenging task given the density of cars and other vehicles as well as the complexity of the environment in urban regions around the world. We describe an accurate and flexible method for detecting vehicles using a template-based directional chamfer matching approach, and vehicle orientation estimation using a refined segmentation followed by a novel Radon transform based profile variance peak detection. The same algorithm was applied to both high resolution satellite imagery and wide area aerial imagery and initial results show robustness to varying illumination and geometric appearance distortions. Nearly 80% of the orientation angle estimates for 1585 vehicles across both satellite and aerial imagery were accurate to within 15◦ of the ground truth. In the case of satellite imagery alone, nearly 90% of the objects had an estimated error within ±1.0◦ of the ground truth.