An Overhead Active-Infrared Vehicle Imaging Sensor Capable of Automated Detection and Classification of Bicycles and Pedestrians
David A. Noyce, Ph.D.
A new algorithm was developed for automated detection and classification of bicycles and pedestrians. Active-Infrared Overhead Vehicle Imaging Sensors equipped with this new algorithm can be used for automated data collection of non-motorized forms of transportation such as bicycles and pedestrians. Automated data collection will be very useful for forecasting future demand for design and policymaking related to these non-motorized forms of transportation. Currently there is no active-infrared device capable of automated detection and classification of bicycles and pedestrians.
- Automated detection and classification of bicycles and pedestrians
- Automated bicycle and pedestrian counts
- Building a reliable database for bicycle and pedestrian applications
- Pedestrian counting at large public events
An existing algorithm uses size and speed to classify motorized vehicles, but does not have the capability to detect and classify bicycles or pedestrians. The new algorithm can accurately classify and count bicycle and pedestrians traveling in either direction on a trail. It is coded in Perl, one of the most portable programming languages, efficient with text processing and string operations. Using the new algorithm associated with the active-infrared sensor, there is a detection accuracy of nearly 100%, and an overall classification accuracy of 92%.
Commercial Ventures and Intellectual Property