Markerless tracking using interest window for augmented reality applications

This paper deals with the problem of object tracking in Augmented Reality (AR) applications. In the recent years AR has been of increasing interest. It consists of inserting a virtual object into a real scene. The temporal coherence between real and virtual objects is one of the hardest challenges of the AR system realization. A markerless tracking based on the interest points matching using SURF (Speed Up Robust Features) is mostly used, but it suffers from high computational complexity. For this reason, we propose a method using an interest window around of the object to be tracked. Moreover, a relationship between the object size and the metric threshold of SURF is investigated. The experimental results show an improvement of the proposed object tracking method performance compared to a standard method based on SURF interest points matching.