基于OCR的自然场景文本识别算法研究

Scholars Journal of Engineering and Technology Pub Date : 2021-11-12 DOI:10.36347/sjet.2021.v09i10.002

Yingchun Zhang

{"title":"基于OCR的自然场景文本识别算法研究","authors":"Yingchun Zhang","doi":"10.36347/sjet.2021.v09i10.002","DOIUrl":null,"url":null,"abstract":"RNN and strengthen the semantic information of the context, BLSTM is used to replace the RNN model for label prediction, and then the CTC algorithm is used to complete the transcription and output the final recognition result. Experimental results show that the improved CRNN text recognition algorithm has an accuracy rate of 96.6%, which is 1% higher than the basic CRNN text recognition algorithm, and this end-to-end network structure design also greatly shortens the text recognition time.","PeriodicalId":379926,"journal":{"name":"Scholars Journal of Engineering and Technology","volume":"27 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-11-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Research of Natural Scene Text Recognition Algorithm Based on OCR\",\"authors\":\"Yingchun Zhang\",\"doi\":\"10.36347/sjet.2021.v09i10.002\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"RNN and strengthen the semantic information of the context, BLSTM is used to replace the RNN model for label prediction, and then the CTC algorithm is used to complete the transcription and output the final recognition result. Experimental results show that the improved CRNN text recognition algorithm has an accuracy rate of 96.6%, which is 1% higher than the basic CRNN text recognition algorithm, and this end-to-end network structure design also greatly shortens the text recognition time.\",\"PeriodicalId\":379926,\"journal\":{\"name\":\"Scholars Journal of Engineering and Technology\",\"volume\":\"27 1\",\"pages\":\"0\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2021-11-12\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Scholars Journal of Engineering and Technology\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.36347/sjet.2021.v09i10.002\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Scholars Journal of Engineering and Technology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.36347/sjet.2021.v09i10.002","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}

引用次数: 0

摘要

RNN和强化上下文的语义信息，用BLSTM代替RNN模型进行标签预测，再用CTC算法完成转录并输出最终的识别结果。实验结果表明，改进的CRNN文本识别算法准确率达到96.6%，比基本的CRNN文本识别算法提高了1%，并且这种端到端网络结构设计也大大缩短了文本识别时间。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

查看原文

微信好友朋友圈 QQ好友复制链接

本刊更多论文

Research of Natural Scene Text Recognition Algorithm Based on OCR

RNN and strengthen the semantic information of the context, BLSTM is used to replace the RNN model for label prediction, and then the CTC algorithm is used to complete the transcription and output the final recognition result. Experimental results show that the improved CRNN text recognition algorithm has an accuracy rate of 96.6%, which is 1% higher than the basic CRNN text recognition algorithm, and this end-to-end network structure design also greatly shortens the text recognition time.

求助全文

通过发布文献求助，成功后即可免费获取论文全文。去求助

来源期刊

Scholars Journal of Engineering and Technology

自引率

0.00%

发文量