Automatic reading of vehicle license plates is considered an approach to mass surveillance. It allows, through the detection / localization and character optical recognition to identify a vehicle in images or image sequences. Many applications such as traffic monitoring, detection of stolen vehicles, the toll or management of parking entrance/exit management using this method.

Yet despite significant progress made since the appearance of the first prototypes in 1979 with a recognition rate sometimes impressive, thanks to advanced scientific research in sensor technology, the constraints imposed for the proper functioning of such systems limit the scope of application. Indeed, optimal use of license plate localization and recognition techniques in operational scenarios requiring controlled lighting conditions and a limitation in the installation, speed, or simply plate?s type. Automatic reading of vehicle license plates then remains an open research problem.

The main contribution of this thesis is threefold. First, a new robust approach to license plate localization in images or image sequences is proposed. Then, improving the quality of localized plates is treated by an adaptation of super-resolution technique. Finally, a unified model of localization and super-resolution is proposed to reduce the time complexity of both combined approaches.