Research On The Automatic Recognition Algorithm Of QR Barcode In Complex Environment

Due to the obvious drawbacks in information capacity, error correction capability and security for1D barcode,2D barcode occurs under this circumstance. It does not need the support of background database. Quick response (QR) barcode originated from its inventor who wished it can be quickly decoded to get its contents. As a common2D barcode, QR barcode has the advantages such as high encoding intensity, less space occupation, fast-speed recognition and strong capability of Chinese character encoding.However, the decoding of2D barcodes including QR barcode has a strong requirement for the pre-processing of captured barcode image and its recognition algorithm. In general, for those2D barcode images captured by scanner, digital camera, especially the camera equipped with mobile phone, there often exist blurring, uneven illumination and various distortions. In addition, there might be partial damages to the barcode itself. These complex factors have serious influence on the correct decoding of QR barcode. After making a brief introduction to the structure of QR barcode and existing work on its automatic recognition, we focus on the key techniques for automatic recognition of QR barcode image. The main work and contribution of this thesis are summarized as follows.First, a low-complexity image preprocessing scheme is proposed for QR barcode recognition by fully utilizing its structural characteristics. It introduces the concept of local window with adaptive position and size to improve the result of binarization. It utilizes the version information to decide the number of corresponding alignment patterns and their rough position. Then, it automatically locates those alignment patterns and adaptively divides barcode image into several regions in terms of these alignment patterns and the three position detection patterns. Thus, the binary bitstreams can be obtained by sampling to each region respectively for automatic recognition. Since it does not utilize the traditional methods such as edge detection and line detection, thus the side effects of background noise and geometric distortion are greatly alleviated.Second, a complete QR barcode software is implemented on the PC platform by modular programming design. It can automatically recognize the contents in QR barcode. Experimental results show that the proposed image pre-processing techniques can realize the automatic recognition of QR barcode under complex situations such as stripe noises, uneven illumination, partial damage and geometric distortion. Moreover, it has a high recognition ratio, and the recognition efficiency can meet the requirement of real-time decoding.