Fully Automated Texture Tracking Based on Natural Features Extraction and Template Matching

Abstract

In this work we propose a novel approach to real-time texture tracking and registration, based on natural feature extraction from planar objects and template matching, Our method is oriented to planar objects with arbitrary textures but with rectangular topologies and well contrasted contours and does not require any external fiducial marker, either for the set-up or the tracking phases. Once the initial pose condition is obtained, previous planar object information is used to compute subsequent planar object’s pose, so that the time coherence of the input video stream is exploited. Our system is completely automated and produces real-time efficient tracking which can be applied to entertainment AR applications and other. The paper discusses also the novelty of the approach, in relation to other existing texture tracking algorithms.