Image Features Detection and Tracking for Image Based Target Augmented Reality Application

Abstract

Image based target used in augmented reality application is not the same as traditional fiducial markers, data markers code and QR-codes in detection and tracking. The available features from the image detected and tracked are naturally found in the image itself. Image target commonly used in recognizing and augmented hard copy media and product for marketing campaigns, gaming, and visualizing products in the environment where the product was targeted to be used. HOwever, for mobile-based augmented reality application, image features detection and tracking still expose many challenges. This paper aims to review previous and current keypoint detection and tracking techniques for image target and address several issues face in mobile based augmented reality application. Most traditional interest-point detection and description methods are handcrafted. These methods focus on interest points in a generic setting and lack provision to adapt toward a specific dataset. Learning-based approaches can be a new solution to real-time application such mobile-based application. This project continuity work from our previous work done for mobile-based augmented reality coloring application [1].