Due to the marked difference in many characteristics between trucks and smaller vehicles, accurate and timely truck data are of significant importance. Unfortunately, few frequent and wide-area truck data are collected with the systems currently in place. Furthermore, the systems that are capable of truck data collection are typically expensive and limited in application. For this reason, wide-area truck data are typically collected every few years, although more timely truck data are desired. There is no doubt that continuous collection of truck data is beneficial to a variety of purposes. This work presents an image processing algorithm for length-based vehicle classification using an image stream captured by an uncalibrated video camera. Although the current implementation separates vehicles based only upon length, the ultimate goal is to develop a system based upon the highway performance monitoring system guidelines. The basis of the algorithm is to relatively compare vehicle lengths to each other to estimate truck volumes and eliminate the need for complicated system calibration. The algorithm was implemented in C#, a new programming language platform developed by the Microsoft Corporation. The system test revealed that the vehicle length classifications estimated by the algorithm do indeed satisfactorily resemble the actual observations. The proposed algorithm may enable the widely installed surveillance video cameras to count classified vehicles including trucks.
Length-based vehicle classification using images from uncalibrated video cameras
2004-01-01
537071 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Transportation Research Record | 2007
|Flexible 3D models from uncalibrated cameras
British Library Online Contents | 1996
|Estimating 3D body pose using uncalibrated cameras
IEEE | 2001
|