Detection of runway boundary in ATC ground radar

The radar environment is usually nonstationary and an efficient adaptation requires real-time exploitation of a priori knowledge concerning the environment surrounding the radar. In an airport contest, the radar scenario is heavily influenced by the inevitable presence of interferences, the background clutter, but other interferences (presence of vehicles and unwanted signs on the runway) contribute to make difficult the target detection. In particular, the paper addresses the detection of the runway boundary in ATC (Air Traffic Control) ground radar signal which can be not well defined due to the grass growth in different period in the islands near the runways. The grass growth can cause critical situation because the signal level of the grass zones is equal to that of the asphalt and it needs a visual analysis by a human operator to determine the presence of objects and targets in this clutter area. An active contour search algorithm (Snake algorithm) is used for detecting the runway boundary in SMR radar images. In our application a standard Snake algorithm is adapted and it is possible to obtain a reasonable approximation of the actual contour, which greatly reduces drawbacks related to the presence of the local minima. The performance of the active contour algorithm is tested on a set of radar images extracted from SMR radar images acquired in Italian airports.