基于Hough变换和深度学习的条形码识别  

Barcode recognition based on Hough transform and deep learning

作　　者：屈源昊 张丰收[1] 昌继宝 丰瑞博 QU Yuanhao;ZHANG Fengshou;CHANG Jibao;FENG Ruibo(School of Mechanical Engineering,Henan University of Science and Technology,Luoyang 471003,China)

机构地区：[1]河南科技大学机电工程学院,河南洛阳471003

出　　处：《现代电子技术》2025年第1期157-162,共6页Modern Electronics Technique

基　　金：国家重点基础研究发展计划(J2019-VII-0017-0159)。

摘　　要：为了解决复杂背景下条形码不易定位识别的难题,提出一种基于Hough变换和深度学习相结合的方法对条形码进行校正定位。首先对待检测图像进行灰度化、高斯模糊以及边缘检测等预处理;然后利用Hough变换检测条形码图像中的线段,进行旋转校正,校正后的图像经Yolov5对条形码进行识别和提取,完成条形码的识别分割。文中方法对不同样式条形码均有较好的识别效果,旋转校正的精确度达到99.31%,识别平均精确度达到99.40%,召回率达到99.79%,推理时间为10.5 ms。提出的方法可对任意角度倾斜进行校正,识别条码具有较高的准确率,对条形码定位识别具有一定的应用价值。In order to cope with the difficulties in the recognition of barcodes in complex background,a recognition method is proposed to carry out the barcode correction and positioning based on the combination of Hough transform and deep learning.The images to be detected are subjected to preprocessing,including gray processing,Gaussian blurring and edge detecting.And then,Hough transform is used to detect the line segments in the barcode image for rotation correction.The barcode image is recognized and extracted by Yolov5 to fulfill the barcode recognition and segmentation.The proposed method has a good recognition effect on different types of barcodes.The accuracy rate of rotation correction of the method is 99.31%,its average accuracy of recognition is 99.40%,its recall rate is 99.79%,and its reasoning time is 10.5 ms.The proposed method can correct any angle inclination,and has high accuracy rate for barcode recognition,so it has a certain application value for barcode location recognition.

关 键 词：HOUGH变换 Yolov5 倾斜校正 条码识别 图像处理 机器视觉 

分 类 号：TN911.73-34[电子电信—通信与信息系统] TP391.41[电子电信—信息与通信工程]