In this study, we follow international standard ISO/IEC 24778 to design and implement, under an industrial vision development platform, a high-performance Aztec 2D barcode decoder with a new calibrationless correction method for non-linear distortion. The proposed correction method horizontally and vertically fits the edges of a non-linear distorted barcode images to two sets of trajectory equations, respectively. References of the trajectory behaviors are those four distorted barcode boundary lines, which were straight before distortion. These best-fit trajectory equations can thereafter be used to reverse the distortion. To simplify the derivation of the sets of the trajectory equations, a reasonable and practical assumption is given: curvature of horizontal or vertical sets of trajectories are changing linearly from top to bottom and from left to right boundaries of the given barcode image. By this assumption, these trajectory equations can be obtained by simple mathematics. In this method, conventional computation-intensive equation solving, e.g. multivariate nonlinear optimizations, is avoided. Thus, an industrial class of barcode reader system with computation-efficient, calibrationless, and real-time non-linear correction can be achieved.
