Moving Vehicle Detection Using a Single Set of Quickbird Imagery — an Initial Study

Moving vehicle detection has become increasingly important in transportation management and road infrastructure development. State-of-the-art moving vehicle detection techniques require multi-frame images or at least a stereo pair of aerial photos for moving information extraction. This paper presents a new technique to detect moving vehicles using just one single set of QuickBird imagery, instead of using airborne multi-frame or stereo images. To find out the moving information of a vehicle, we utilize the unnoticeable time delay between multspectral and panchromatic bands of the QuickBird (or Ikonos) imagery. To achieve an acceptable accuracy of speed and location information, we first employ a refined satellite geometric sensor model to precisely register QuickBird multispectral and panchromatic bands, and then use a new mathematic model developed in our research to precisely calculate the ground position and moving speed of vehicles detected in the image. The initial testing results demonstrate that the technique developed in this research can extract information on position, speed, and moving direction of a vehicle at a reasonable accuracy.