i (b) (c) (01)12345678901231 (d) Barcode Localization With Region-Based Gradient Statistical Analysis

Barcode, as a kind of data representation method, has been adopted in a wide range of areas. Especially with the rise of the smart phone and the hand-held device equipped with high resolution camera and great computation power, barcode technique has found itself more extensive applications. In industrial field, barcode reading system is highly demanded to be robust to blur, illumination change, pitch, rotation, and scale change. This paper gives a new idea in localizing barcode under a region-based gradient statistical analysis. Making this idea as the basis, four algorithms have been developed for dealing with Linear, PDF417, Stacked 1D1D and Stacked 1D2D barcodes respectively. After being evaluated on our challenging dataset with more than 17000 images, the result shows that our methods can achieve an average localization accuracy of 82.17 % with respect to 8 kinds of distortions and within an average time of 12 ms