Novel algorithm of vehicle detection by using new ultra resolution aerial image, Three Line Scanner

Traffic congestion on roads in cities such as Tokyo is becoming worse. Therefore, collecting on-street vehicle statistics is important for traffic management in the short or long term. Therefore, a new approach to vehicle detection using the Three Line Scanner or TLS, and new imagery has been proposed. The framework of the approach consists of four steps: preparation, vehicle detection, moving/stopping vehicle discrimination and parking/signals waiting vehicle classification. This study focuses on the first two steps of a study framework. Pre-processing is the approach to prepare fundamental information for further processing such as image segmentation, non-vehicle removal, building shadow correction and so on. The second stage is the algorithm of vehicle detection using a vehicle explicit model on ultra-resolution images of TLS. The third stage involves the newly-proposed concepts of moving/parking discrimination by applying stereoscopic measurement. The final stage is the new approach of parking/signal waiting classification by integrating a digital road map. The first two stages of a framework have been fulfilled with promising results. The third stage has been proposed conceptually and the last stage is being developed with conceptual design.