The use of shadows to identify automatically vehicles in urban areas

• Autor(a) principal: Santos, Daniel Rodrigues dos
• Data de Publicação: 2008
• Outros Autores: Dalmolin, Quintino, Basso, Marcos Aurélio, Araki, Hideo
• Tipo de documento: Artigo
• Idioma: por
• Título da fonte: Boletim de Ciências Geodésicas
• Texto Completo: https://revistas.ufpr.br/bcg/article/view/11247
• topic: Veículos; Reconhecimento Automático; Imagem de Intensidade Laser; Modelo Implícito; Processamento Digital de Imagens; Vehicles; Automatic Recognition; Laser Intensity Image; Implicit Model; Digital Image Processing

In this paper we proposed an automatic methodology to identify vehicles presents in aerial digital images. The vehicles are identified by using color attributes and a new proposed implicit model. The developed mathematical model, searches for pixels showing the lowest digital numbers intensity, such as that describes which are associated with the shadows of cast by vehicles. In the proposed methodology, shadows are used as heuristic information, to help identifying vehicles present on aerial image. A laser intensity image of region is used to reduce the searching space to the local roads, optimizing the required computing time in this way. This approach allows the identificationand the labeling of the vehicles. The goal in this work is to identifyautomatically vehicles present in aerial image by using the shadows cast on them. The experiments have shown that the proposed model yielded accurate results, proving to be capable of performing with reasonable accuracy the automatic detection of vehicles in digital image data.