Hybrid SURF-golay marker detection method for augmented reality applications

Abstract

Augmented reality is a visualization technique widely used in many applications including different design tools. These tools are frequently based on tracking artificial objects such as square markers. The markers allow users to add a 3D model into the scene and adjust its position and orientation. Nevertheless, there are significant problems with marker occlusions caused by users or objects within the scene. The occlusions usually cause a disappearance of the 3D model. Such behavior has substantial negative impact on the application usability. In this article we present a hybrid marker detection approach. With this approach, markers are detected using the well-known SURF method. This method is able to recognize complex natural objects and deal with partial occlusions. Further, we overcome the problem of distinguishing similar markers by using the Golay error correction code patterns. The described approach represents a robust method that is able to identify even significantly occluded markers, differentiate similar markers, and it works in a constant time regardless of the amount of used markers.