Skip to Main Content
Multi-target tracking is a crucial but complex problem in small target tracking applications, especially when large number of false alarms existed. In order to simplify the multi-target tracking, a novel simple algorithm is presented by using motion properties to reduce false alarms, which is implemented through mathematical morphology and multi-frame cumulation. Firstly, the consecutive images are dilated and cumulated according to the motion properties of the small targets, thus generate the cumulating image. Secondly, most of the false alarms are reduced from the cumulating image through mathematical morphological operation. Finally, the resulting consecutive images including targets and very few false alarms are restored from the processed cumulating image and the original consecutive binary images. Experiment results illustrated that most of the false alarms were excluded and the small targets were easier to be identified, which largely decreased the computational time of target tracking algorithm, and thereby decreasing the complexity of the small targets tracking.
Date of Conference: 3-5 June 2008