Identification of QR codes based on pattern recognition

To avoid the limitation of present QR code algorithms which are only available on the paper presswork, this paper proposed an algorithm based on Pattern Recognition to realize the identification of the QR codes printed on various materials with different models. The QR codes from the sample images were manually marked first, and then the input images were divided into blocks. For every single block, its MRH and LBP features were calculated and such texture patterns were trained to obtain a QR code identifier using the Spatial Boost algorithm. For real-time identification, this identifier could divide the real-time input images, and output the classified results of the blocks, considering the texture features (MRH, LBP) of the blocks as input vectors. The blocks confirmed to be 2D barcode were combined into QR codes. The QR codes identifier obtained by the last training reached a 100% rate of detection in the test database of the experiment.