Desenvolvimento de metodologia para monitoramento remoto de rodovias: VANTRod

• Ano de defesa: 2017
• Autor(a) principal: Di Renzo, André Biffe
• Tipo de documento: Dissertação
• Tipo de acesso: Acesso aberto
• Idioma: por
• Instituição de defesa: Universidade Tecnológica Federal do Paraná; Curitiba; Brasil; Programa de Pós-Graduação em Engenharia Elétrica e Informática Industrial; UTFPR
• Palavras-chave em Português: Rodovias; Processamento de sinais - Técnicas digitais; Algorítmos computacionais; Rodovias - Medidas de segurança; Engenharia elétrica; Roads; Signal processing - Digital technique; Computer algorithms; CNPQ::ENGENHARIAS::ENGENHARIA ELETRICA::TELECOMUNICACOES::SISTEMAS DE TELECOMUNICACOES; Engenharia Elétrica
• Link de acesso: http://repositorio.utfpr.edu.br/jspui/handle/1/2852

Highways are the principal transportation modal way in Brazil for cargo and passengers. Highways can suffer wear due to weather and traffic load, hence necessary a constant monitoring of the pavement and traffic signalization health. In general, the highway health is monitored manually being necessary persons to make the verification process. One alternative to this process is the use of aerial images. This work presents a highway remote sensing methodology from aerial imagens and digital image processing (DIP), as a tool to verify the road conditions, including the pavement and road markings. The acquisition of images are performed by an Unmanned Aerial Vehicle (UAV) enabling large area scans with less time. With the acquired imagens, DIP techniques and pattern recognition are employed to extract and identify highways parameters. So, an algorithm was developed to process and extract road information of the acquired images providing inspection of highway with agility and precision. The developed algorithm has three parts: the first one make the roadway segmentation, the second segments objects of the road and the third classifies the segmented objects. To classify the segmented objects, the Histogram of Oriented Gradient (HOG) descriptor has been used to extract characteristics of the objects and the Support Vector Machine () was used to classify the objects. With this developed algorithm, positive results has reached in obtain road information from the aerial images. Performance tests has been performed and a hit rate of 97.37% was reached for the selected classes, proving the ability of this proposed methodology could be applied in real environment helping maintenance and management highway teams.