Studies On The High-speed Recognition Algorithms For 2-D Barcodes

Studied in this thesis are a set of fast recognition algorithms for a 2D barcode, Maxicode, existing in a big image. Those algorithms are related to three recognition stages, i.e., location, decoding, and error correction.The algorithms proposed in this thesis are based on image processing techniques. After an image is captured via camera system, it is processed with the algorithms on PC to perform image analysis to detect the locations of barcodes and decode them. Such a image based approach is most suitable for recognizing 2D barcodes, especially matrix symbols, which can be found extensive applications in automations.Proposed in this thesis is a new accurate Maxicode boundary location algoritm, which is based on edge detection. It takes advantages of Maxicode inherent characteristics, including its Finder Pattern, Orientation Patterns, and quiet zones, to locate the barcodes in an image with relatively simple computations. No time-consuming complicated image processing algorithms, such as Fourier transformation, image rotation (coordinate conversion), momentum invariant computations, et al., are needed. After barcodes are located, they are decoded according to Maxicode's encoding rules. The messages decoded include both normal message and structured carrier message. In order to deal with any potential damages to the barcodes and other possible errors, a Peterson-Gorenstein-Zierler error correction algorithm is employed to find and correct any errors in the decoded message.The complete recognition procedure is simple and easy to understand. Its image processing algorithms are robust and time efficient, thus can meet real-time requirements in real-life industry applicatons. Test runs show that the algorithms are fast and not sensitive to image and barcode size or orientation.