Skip to Content

HMPMR Strategy for Real-Time Tracking in Aerial Images, Using Direct Methods

Publication
Journal: 
Machine Vision and Applications

0932-8092

Publication Date: 
2014

Abstract:
The vast majority of approaches make use of features to track objects. In this paper we address the tracking problem with a tracking-by-registration strategy based on direct methods. We proposes a hierarchical strategy in terms of image resolution and number of parameters estimated in each resolution, that allows direct methods to be applied in demanding realtime visual tracking applications. We have called this strategy the Hierarchical Multi-Parametric and Multi-Resolution strategy (HMPMR). The Inverse Composition Image Alignment Algorithm (ICIA) is used as image registration technique and it is extended to an HMPMR-ICIA. The propose strategy is tested with different datasets; and also with image data from real flight tests using an Unmanned Aerial Vehicle (UAV), where the requirements of direct methods are easily unsatisfied (e.g. vehicle vibrations). Results show that by using an HMPMR approach, it is possible to cope with the efficiency problem and with the small motion constraint of direct methods, conducting the tracking task at real-time frame rates and obtaining a performance that is comparable to, or even better, than the one obtained