This paper describes an image processing algorithm for pedestrian detection and tracking in infrared images. The algorithm structure combines a straight forward with a backward loop methodology to combine fast initiation functions with more affordable verification functions. To create reliable measurements at the initiation level a two-step blob detection and filtering algorithm is performed. A new adaptive threshold operator is introduced which uses the Entropy value of the whole image to get optimal threshold results of infrared images. At the blob filtering step a new formulation for fast and accurate computation of geometric moments is presented. The developed verification method uses a fuzzy logical combination to evaluate the results of the head and torso matching. Finally, a multi target capable Kalman filtering is used in order to track the recognized pedestrians. The developed algorithm has been implemented on an experimental vehicle equipped with an infrared camera and was successfully tested.
A new approach for pedestrian detection and tracking in infrared images
Ein neues Näherungsverfahren zur Erkennung und Verfolgung von Fußgängern mit infraroten Bildern
2005
5 Seiten, 15 Bilder, 7 Quellen
Conference paper
English
Pedestrian Detection and Tracking in Far Infrared Images
British Library Conference Proceedings | 2005
|A new Approach for Pedestrian Detection and Tracking in Infrared Images
British Library Conference Proceedings | 2005
|Pedestrian detection in infrared images
IEEE | 2003
|Shape and Motion -based Pedestrian Detection in Infrared Images: A Multi Sensor Approach
British Library Conference Proceedings | 2005
|Pedestrian tracking using thermal infrared imaging
SPIE | 2006
|