Poster: a coarse-fine corner detection approach for two-dimensional barcode decoding

Two-dimensional barcodes are widely used in mobile advertisement business, while their decoding performance is not always satisfactory under uncontrolled environments. The corner detection accuracy has been identified as a critical factor affecting the overall system performance. The standard barcode detector performs a candidate search in the binarized barcode image based on the rectangular shape of the barcode. Its performance is not very accurate due to the limited accuracy of the binarized image. In this work, we proposed a coarse-fine corner detection approach for locating the barcode region. It performs far more accurately than the standard barcode detection scheme while keeps the computational complexity affordable. Experimental results for high capacity barcodes show that the proposed detection scheme can extend the range of operation parameters, such as wider angles, and much lower the detection bit error rate, relative to the standard barcode decoder.