基于Qt与Arm NN的嵌入式喷码检测系统设计与实现
投稿时间:2019-04-28  修订日期:2019-05-05  点此下载全文
引用本文:
摘要点击次数: 207
全文下载次数: 0
作者单位邮编
王斌 广西师范大学 541004
宋树祥* 广西师范大学 541004
王宜瑜 广西师范大学 
庞中秋 广西师范大学 
基金项目:广西创新驱动重大专项:专业相机手持稳定器关键技术研发及产业化 项目编号:桂AA18118009
中文摘要:现存的喷码检测系统大多基于工控PC电脑实现,其在价格、体积、功耗、灵活性等方面存在着较大的局限性。针对这一问题,设计了一种基于Qt与Arm NN的嵌入式喷码检测系统,该系统的核心处理器采用Arm cortex-a9,搭载配置有相关软件工具的嵌入式linux操作系统,采用位置检测算法来实现软件检测待测目标的位置,通过由图像预处理、喷码区域提取、条形码识别、字符子串分割、基于卷积神经网络及caffe框架的喷码字符识别模型训练、利用Arm NN将识别模型引入到ARM端实现喷码字符识别、识别结果处理等过程组成的喷码检测算法来实现对喷码内容的检测。该系统还基于Qt多线程技术开发了具有完善系统功能的图形应用程序,便于现场生产人员使用。经相关测试表明该系统有效可行、正检率高,能够满足包装生产流水线喷码检测的实际需求。
中文关键词:嵌入式  区域提取  字符分割  卷积神经网络  Arm NN  字符识别
 
Design and implementation of embedded printing character inspection system based on Qt and Arm NN
Abstract:Most of the existing printing character inspection systems are based on industrial PC, but they are limited in price, volume, power consumption and flexibility.To solve this problem, an embedded printing character inspection system based on Qt and Arm NN is designed. The core processor of the system is Arm cortex-a9, which is equipped with embedded Linux operating system with relevant software tools. The position detection algorithm is adopted to realize the location of the target to be tested.By image preprocessing, region extraction, bar code identification, character substring, based on convolutional neural network segmentation and caffe framework of character recognition model training, using Arm NN identification model is introduced into the Arm end realize the recognition of character recognition, the results of treatment process, such as printing inspection algorithm to implement to spurt the code content detection.The system also developed a graphic application program with perfect system function based on Qt multi-thread technology, which is convenient for field production personnel.Relevant tests show that the system is effective and feasible, with high positive inspection rate, and can meet the actual needs of the packaging production line for the printing character inspection.
keywords:Embedded  Area extraction  Character segmentation  Convolutional neural network  Arm NN  Character recognition
查看全文   查看/发表评论   下载pdf阅读器