Detecção de placas de licenciamento veicular utilizando imagens com diferentes resoluções e em ambiente não controlado empregando técnicas de super resolução

Abstract

In this dissertation, we propose a solution using Digital Image Processing techniques to detect license plates of Brazilian vehicles, which consists in elaborating a robust and flexible method in relation to the complex textit background with the acquisition made in motion to perform the detection of Vehicle license plates. Detection occurs automatically employing PDI techniques, which include spatial domain filters and to help us, we use the opentext library OPENCV in C/C++. In addition, as one of the contributions, we approached the use of the T V − L1 Bilateral Super Resolution algorithm in the database videos in order to contribute to the detection of the object of interest. To perform the experiments, we used a database that was designed containing videos with three different resolutions, the acquisition was performed with camera and object of interest both moving with a complex textit background, where, from the videos, we employed sequential images in the experiments performed. The results of the experiments covered two scenarios, a first scenario was not employed the Super Resolution algorithm and a second, we employed the Super Resolution algorithm, thus, it was possible to validate the proposed methodology by demonstrating the ability to detect license plates on the challenge. However, we propose that the standardization of image resolutions may influence the results.