期刊文献+

改进SSD算法的QR二维码精准定位方法研究

Research on Accurate QR Two-Dimensional Code Localization Method with Improved SSD Algorithm
下载PDF
导出
摘要 为解决污水处理厂加盖封闭后带来的巡检难题,可利用巡检机器人精准定位标识有特征二维码的检测盖板,并打开盖板以获取被巡检区域内的关键信息。由于传统二维码检测技术无法满足高精度的定位需求,提出一种改进的单次多边框检测(SSD)算法。在SSD算法对快速响应(QR)二维码区域进行粗定位的基础上,通过中值滤波和自适应二值化处理,采用透视变换方法矫正变形图像,利用边缘提取和定位码筛选的方法实现对QR二维码的精准定位。构建污水处理厂检测盖板QR二维码数据集,选取图像采集中常见的5种工况进行试验。试验结果表明,相较于开源计算机视觉库(OpenCV)算法,改进SSD算法识别率平均提升约5%、开盖率平均提升约32%,大幅提高了智能巡检效率。该算法有效解决了复杂工业场景下的二维码图像定位问题。 In order to solve the inspection problem brought about by the closure of wastewater treatment plants with cover plate,inspection robots can be utilized to accurately locate the inspection cover plate marked with characteristic two-dimensional codes,and open the cover plate in order to obtain the key information in the inspected area.Since the traditional two-dimensional code detection technology can not meet the demand for high-precision localization,an improved single shot multibox detector(SSD)algorithm is proposed.On the basis of the SSD algorithm’s coarse positioning of the quick response(QR)two-dimensional code area,the median filtering and adaptive binarization are used to correct the deformed image by using the perspective transformation method,and the accurate positioning of the QR two-dimensional code is realized by using the edge extraction and positioning code screening method.The QR two-dimensional code dataset of wastewater treatment plant inspection cover plate is constructed,and five common working conditions in image acquisition are selected for testing.The experimental results show that compared with the open-source computer vision library(OpenCV)algorithm,the improved SSD algorithm increases the recognition rate by about 5%on average,and the cover opening rate by about 32%on average,which significantly improves the efficiency of intelligent inspection.The algorithm effectively solves the problem of two-dimensional code image localization in complex industrial scenarios.
作者 袁征 韩奕 YUAN Zheng;HAN Yi(Shanghai Municipal Engineering Design Institute(Group)Co.,Ltd.,Shanghai 200092,China)
出处 《自动化仪表》 CAS 2024年第12期34-38,共5页 Process Automation Instrumentation
关键词 污水处理厂 智能巡检机器人 快速响应二维码 目标检测 精准定位 深度学习 单次多边框检测算法 Wastewater treatment plant Intelligent inspection robot Quick response(QR)two-dimensional code Target detection Precision localization Deep learning Single shot multibox detector(SSD)algorithm
  • 相关文献

相关作者

内容加载中请稍等...

相关机构

内容加载中请稍等...

相关主题

内容加载中请稍等...

浏览历史

内容加载中请稍等...
;
使用帮助 返回顶部