期刊文献+

结合空洞卷积和先验阈值的数显仪表识别

Digital Display Instrument Based on Dilated Convolution and Prior Threshold Detection
下载PDF
导出
摘要 数显仪表在工业生产中应用广泛,目前,数显仪表的自动化识别主要基于传统图像处理。针对传统图像处理易受环境影响,处理步骤繁杂,准确率低等问题,论文提出了一种基于深度学习的自动化数显仪表读数方法。首先,使用Pixellink字符检测算法对数显仪表进行粗检测,根据实际图像设置裁剪阈值,使用结合空洞卷积的CRNN架构完成对裁剪切片的识别。训练过程中,借鉴了迁移学习的思想,加载了在RCTW数据集上的预训练模型。相比于基于字符分割的传统识别方法,论文方法省去了复杂的预处理与后处理,实现了端到端的数显仪表识别。以变电站数显仪表为例,所提方法达到了99.3%的准确性,具有一定的鲁棒性与实用性。 Digital display instrument is widely used in industrial production.At present,automatic recognition of digital display instrument is mainly based on traditional image processing.Traditional image processing is easy to be affected by the environment,the processing steps are complex,and the accuracy is low.In this paper,an automatic digital instrument reading method based on deep learning is proposed.Firstly,Pixellink character detection algorithm is used for coarse detection of digital instrument,clipping threshold is set according to the actual image,and CRNN architecture combined with hole convolution is used to complete the recognition of clipping.In the training process,the idea of transfer learning is used for reference,and the pre training model on RCTW dataset is loaded.Compared with the traditional recognition method based on character segmentation,this method saves the complex pre-processing and post-processing,and realizes the end-to-end digital instrument recognition.Taking the digital display instrument in substation as an example,the proposed method achieves 99.3%accuracy and has certain robustness and practicability.
作者 封磊 李晓明 FENG Lei;LI Xiaoming(School of Computer Science and Technology,Taiyuan University of Science and Technology,Taiyuan 030024)
出处 《计算机与数字工程》 2023年第4期855-859,共5页 Computer & Digital Engineering
基金 国家自然科学基金项目(编号:61373099)资助。
关键词 数显仪表识别 空洞卷积 先验阈值 端到端 digital display instrument dilated convolution prior threshold end-to-end
  • 相关文献

参考文献6

二级参考文献33

  • 1张海波,段会川,张曙光,刘弘.一种数字仪表显示值快速识别方法[J].计算机工程与应用,2005,41(4):223-226. 被引量:26
  • 2张海波,段会川,谢术富,刘弘.一种数字仪表显示值识别的预处理算法[J].计算机应用研究,2005,22(2):240-242. 被引量:9
  • 3张宏烈.支持向量机在字符识别中的应用研究[J].微计算机信息,2006(04Z):245-247. 被引量:11
  • 4ANIL K. JAIN,BIN YU.AUTOMATIC TEXT LOCATION IN IMAGES AND VIDEO FRAMES[J]. Pattern Recognition . 1998 (12)
  • 5Huiping Li,David Doermann,Omid Kia.Automatic text detection and tracking in digital video. IEEE Transactions on Image Processing . 2000
  • 6Qixiang Ye,Qingming Huang,Wen Gao,Debin Zhao.Fast and robust text detection in images and video frames[J]. Image and Vision Computing . 2005 (6)
  • 7Yu Zhong,Kalle Karu,Anil K. Jain.Locating text in complex color images[J]. Pattern Recognition . 1995 (10)
  • 8R. Lienhart,A. Wernicke.Localizing and segmenting text in images and videos. IEEE Transactions on Circuits and Systems for Video Technology . 2002
  • 9Huiping Li,Doermann D.A video text detection system based on automatedtraining. Pattern Recognition,2000.Proceedings.15th International Conference onVolume 2 . 2000
  • 10Q.Song,M.Cai,M.R.Lyu.A new approach for video text detection. Proc International Conference on Pattern Recognition . 2002

共引文献86

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