An adaptive background-learning and subtraction method is proposed and applied to two real-life traffic video sequences to obtain more accurate spatiotemporal information on the vehicle objects. When paired with the image segmentation, the method is robust under many conditions. A key advantage of the method is that it is fully automated and unsupervised; it performs the generation of background images using a self-triggered mechanism.
- Intelligent transportation systems
- Robotic vision
- Vehicle tracking
- Video analysis
ASJC Scopus subject areas
- Control and Systems Engineering