J. Martí et al. (Eds.): IbPRIA 2007, Part I, LNCS 4477, pp. 137–144, 2007. © Springer-Verlag Berlin Heidelberg 2007 Dealing with the Perspective Distortion to Detect Overtaking Cars for Driving Assistance S. Mota 1 , E. Ros 2 , J. Díaz 2 , R. Agís 2 , R. Rodriguez 2 , and R. Carrillo 2 1 Department of Computer Sciences and Numerical Analysis, University of Córdoba, Campus de Rabanales s/n, Edificio Albert Einstein, 14071, Córdoba, Spain 2 Department of Computer Architecture and Technology, University of Granada, Periodista Daniel Saucedo Aranda s/n, 18071 Granada, Spain smota@uco.es, {eros,jdiaz,ragis,rrodriguez,rcarrillo}@atc.ugr.es Abstract. The driver’s loss of attention is an important problem in which are spent considerable research efforts in different areas such as psychology, automobile technology, computer vision and driving assistance. We use here a simple algorithm based on rigid-body and motion detection. This scheme efficiently segments moving objects using the visual field of the driver’s rear- view mirror. The overtaking scene in the rear-view mirror is distorted due to perspective, making it difficult to detect the overtaking car. Thus we propose two alternative methods to deal with this problem and compare the results in different overtaking sequences. 1 Introduction Biological systems represent high efficient computing schemes. In particular, visual motion detection is complex and accurate system. It is expected that in 10 years, vehicles will incorporate devices based in computer vision applied to driving assistance. The long-medium term goal is to implement devices based on vertebrates’ visual systems, because of their amazing efficiency in analysing dynamic scenes. However, current vision models require high computational resources. Therefore a system for driving assistance working in real time might be simpler if designed specifically for this task. As an option it can be used motion detection scheme based on insects motion detection systems, which is a good example of a simple visual motion detection model. Reichardt described a correlation model of motion detection based on elementary motion detectors (EMDs) that emulate the behaviour of early visual stages in insects [1]. In this work, we use an algorithm which in its initial stage is a motion-detection algorithm (based on EMDs) and in a second stage includes rigid-body-motion (RBM) rules to filter noisy patterns. We apply this approach to overtaking scenarios, which is one of the most dangerous situations in driving because the rear-view mirror is sometimes not watched by the driver or is momentarily useless because of the blind spot. The whole algorithm detects the moving vehicle behind and determines whether it is approaching