The use of shadows to identify automatically vehicles in urban areas

Abstract

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 identification
and the labeling of the vehicles. The goal in this work is to identify
automatically 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.