Automatic recognition of airfield runways based on Radon transform and hypothesis testing in SAR images

摘要

A novel approach to recognize the airport runway in SAR image based on Radon transform and hypothesis testing was proposed in this paper. Firstly, a nonlinear edge detection method with arithmetic average and geometric average (A/G) coefficient was adopted to extract the edges of airfield runway. Secondly, Radon transform was performed on the edge image. Several max values of Radon transform matrix were reserved, while the rest were set to zero. Thirdly, with inverse Radon transformation, the Radon transform matrix was transformed to a new binary image, where the main straight lines of runways were retained and other noisy lines and region of no interest were excluded. Finally, the airfield runway was identified by hypothesis testing with the knowledge of the runway's intensity property and structure features. Experimental results demonstrated the proposed method could automatically recognize the airfield runway effectively.