2D mobile barcode

• Main Author: Li, Zheng Jie
• Other Authors: Kot Chichung, Alex
• Format: Final Year Project
• Language: English
• Published: 2010
• Subjects: DRNTU::Engineering::Electrical and electronic engineering::Applications of electronics
• Online Access: http://hdl.handle.net/10356/40335
• Institution: Nanyang Technological University

This report presents a novel five-step framework for 2D Mobile Barcode recognition. The framework consists of Barcode Binarization, Barcode Localization, Barcode Geometry Correction, Barcode Pattern Estimation and Barcode Error Recovery. New algorithms are proposed forbarcode binarization, barcode localization and pattern estimation. Barcode Geometry Correction adopts the traditional inverse perspective transformation. Error Recovery is left for future work. Several difficulties are identified in the barcode recognition processes. Firstly, mobile-captured images are usually of poor quality. Noise, blurriness and irregular illumination are introduced when capturing barcode images. All these undesired conditions add difficulties to image binarization. In order to address this problem, novel binarization algorithm is proposed to efficiently and more accuratelybinarize barcode images. Secondly, captured barcode may locate anywhere on the image. Image noise is another hurdle for the barcode localization process. To solve this issue, novel barcode localization algorithm is proposed to locate barcode accurately, which is experimentally proved to be robust to noises. Thirdly, because of the poor quality of captured image, barcode codeword boundaries are curved. Pattern estimation method is also proposed to divide the barcode into smallest data units–codewords, which are essential for barcode decoding process. Experiments show 99.2% accuracy for the new proposed barcode localization algorithm and 30% overall decoding success rate improvement.