Detection of divided planar object for augmented reality applications

In this research study, we propose a divided planar-object detection method for augmented reality(AR) applications. The proposed system prepares a database of the target object´s natural features, and applies progressive sample consensus (PROSAC), which is a robust estimation method, for iterative homography calculation to achieve the multiple planar-object detection. Moreover, the proposed method can detect shapes of pieces by simultaneously using an occlusion detection method. We demonstrate that it is possible to interact with an arbitrarily divided planar object in real time by our method to implement some AR applications.