Building an Augumented Reality-Based Product Promotion System with ARToolkit Integrated with an Adaboosted Classifier-Assisted 1-D Barcode Reader

The augmented-reality (AR) applications built using unmodified ARToolkit are limited in that the marker patterns and 3D models must be built into the applications themselves, and so are limited in number. We propose a solution to bypass this limitation by integrating ARToolkit with a 1-D barcode reader assisted by an Adaboosted classifier and accessibility to the internet. In this way, marker patterns and 3D model data can be stored on a remote data server and accessed by using UPC barcodes as key. In this way, the marker patterns are decoupled from the applications, and the number of useable marker patterns and 3D model data at runtime becomes unlimited. In addition, we used the adaboost algorithm provided by OpenCV to train a classifier to recognition 1-D barcode in order to quickly isolate the barcode and help increase accuracy of decoding. Several modifications were made to make this system more user-friendly. One of these modifications is the use colorful markers rather than the traditional black-and-white markers used in most ARToolkit applications. However, not all colored markers can be used as stable markers. Experiments were performed to isolate the stable markers, and the results are presented.