Phase-correlation-based Moving Vehicle Velocity Determination from Single-pass WorldView-2 Imagery
Abstract
A new method has been developed for detecting vehicles and estimating their velocities using a single optical very-high-resolution (VHR) image. The method is based on the fact that there is a slight time lag between the acquisitions of data by the two multispectral (MS) sensors on the same VHR platform. Thus, the speed of a moving vehicle can be determined from the difference in the positions of the vehicle observed in the MS images due to this time lag. First, a threshold approach can be used to extract a vehicle from the MS PC1 sub-image whose resolution is 0.5 m. Then, an area-correlation method (i.e., normalized phase correlation) is used to determine the difference in the location of a vehicle from the MS image pair at sub-pixel level. Finally, the velocity of the moving vehicle is calculated based on its displacement. Experimental results indicated that the velocities of moving vehicles on the North Fifth Ring Road of Beijing were within the speed limit (100 km/h).
Keywords
Phase-Correlation, Vehicle, WorldView-2
Publication Date
DOI
10.12783/dtetr/iect2016/3741
10.12783/dtetr/iect2016/3741
Refbacks
- There are currently no refbacks.