High-speed Stable Identification Of 2D Barcode In Complex Scenes

As a kind of high information integration and strong anti-jamming capability,2D barcodes are increasingly used in mobile terminals and industrial assembly lines.Among them,QR code is commonly used in various types of mobile terminals because of unique positioning characteristics and quick response characteristics.Data Matrix code is often used for identification of small industrial production or vulnerable product logo preservation due to its high information density,small size,and strong anti-fouling ability.With the rapid development of two-dimensional codes,the application scene is becoming more and more complex,The traditional two-dimensional code recognition has been unable to meet the requirements of industrial production lines.Therefore,an efficient and stable 2D barcodes identification algorithm for complex scenes has an urgent market demand.The key technology of 2D barcodes recognition mainly lies in three aspects:image preprocessing,2D barcodes positioning and correction.This article puts forward an optimization method based on traditional methods adapt to common industrial complex scenes and efficiency requirements:? Study Sauvola's local adaptive binarization method and Propose an improved Sauvola algorithm based on adaptive window size which adapts to uneven light industrial scenes.? According to features of position detection pattern,special width streams of 1:1:3:1:1 are detected by non-maximally suppressed edge scanning and then clustering them to obtain the precise positions of three“?”characters.It can solve perspective and most of the occlusion issues.? Analyze the edge features of 2D barcodes and use RANSAC iterative fitting to accurately locate four vertices.Interval sampling and removal of non-edge points by the slope are used to realize more efficient positioning algorithm.? According to the distribution characteristics of DM symbols,connected region labeling algorithm based on binary image is used to extract candidate regions of 2D barcodes.Stable segmentation of 2D barcodes and backgrounds in complex scenes is achieved.? A coarse L-shape boundary location method based on projection integration is proposed to improve the efficiency and stability of Hough transform,which improves the positioning efficiency of DM code.This paper realizes efficient identification of 2D barcodes in complex scenes by existing recognition algorithm with above improved methods,It compares existing mature libraries and platform tools,the result demonstrates the innovative algorithm proposed in this paper is more effective.