Automatic Video System for Aircraft Identification J.M. Molina, J. García Dpto. Informática Universidad Carlos III de Madrid. Leganés, Madrid, SPAIN molina@ia.uc3m.es jgarcia@inf.uc3m.es A. Berlanga, J. Besada, J.Portillo GPSS-DSSR- ETSIT Universidad Politécnica de Madrid Madrid, SPAIN aberlanga@grpss.ssr.upm.es besada@grpss.ssr.upm.es portillo@grpss.ssr.upm.es Abstract - Advanced Surface Movement Guidance and Control Systems need the identification of aircraft and vehicles in airport movement area. In this work, a video identification algorithm, based on the tail number recognition, is proposed as a part of a global surveillance video system. The aircraft identification problem has to deal with three fundamental aspects: capturing and pre- processing the images, the international regulation defining the tail number grammar and the pattern recognition methodology. Considering these aspects, the developed system is based basically on three ideas: in the first place, to use the local grey level contrast to detect the characters of the tail number; in second place, to find the zone where the tail number is written to isolate it of the rest of the image; and finally to process this zone to identify the aircraft, using an aircraft database. Keywords: ASMGCS, Video Surveillance, Aircraft Identification, Image Processing, Pattern Recognition. 1 Introduction Advanced Surface Movement Guidance and Control Systems (A-SMGCS) [1][2][3] requires the unambiguous identification of all aircraft and vehicles in the airport movement area. The sensors used currently (or in the near future) to perform airport surveillance are Surface Movement Radar (SMR)[4], Multilateration systems (MS)[5][6], differential GPS broadcasted through a digital data-link (DGPS) and, finally, Video cameras (as described in this paper). The developed identification system is part of on-going VICTOR (“Visualización Integrada para Control de TORre” - Integrated Visualisation for Tower Control) project, launched by AENA (Spanish Air Navigation Services Provider) in 1997 as the first step towards the implementation of an A-SMGCS for Madrid/Barajas Airport. Nevertheless, one of the main requisites in this design is that it should be easily deployed in other airports (AENA currently manages all Spanish airports). Video cameras are a non-cooperative sensor, able to provide both identification and tracking avoiding the need of installing additional avionics on board [7]. The method used to obtain target identification for aircraft is performing an image based tail number recognition, through an optical character recognition (OCR) algorithm. The only restriction of TV cameras is the necessity of clear meteorological conditions that means not too dense fog, rain or snow. Occlusions and operation on bad weather conditions are the main problems of this system. The identification system described here is part of a global system that provides controllers (and potentially pilots) with a display of the location of all surface traffic [8], enabling its separation and guidance in all types of weather conditions without reducing the number of operations or the level of safety. The whole image-based surveillance system will be designed to perform accurate detection and position measurements of aircraft and vehicles on all airport areas and 2D multitarget tracking [9]. The implemented tracking system [10] is based on a distributed tracking architecture, in which every camera has a dedicated positioning/tracking system which calculates target trajectories (local tracks) in the camera projected plane. The approach used to obtain aircraft identification has been to use an OCR to recognize aircraft tail number [11]. Cameras for this function should be deployed near the taxiways and runways, in positions being traversed by all the interest targets, prior to their entrance in the area to be controlled (mainly runways and taxiways). When an aircraft passes in front of the camera (which may be predicted using the tracking system described before), an image of its tail is captured. Then, the aircraft identification algorithm is applied. The problem of aircraft tail number identification in an airport, by means of tail number recognition, comprises three fundamental aspects: how to capture and pre-process the images to identify the tail number, how to use the international regulation that defines the grammar to 1387 ISIF © 2002