融合CTPN和CRNN对识别影像图片文字及应用的研究  

Fusion of CTPN and CRNN for recognition text in image pictures and applications

作　　者：石志强 周新辉 沈康畅 石美静 丁琳[1] Shi Zhiqiang;Zhou Xinhui;Shen Kangchang;Shi Meijing;Ding Lin(School of Computer Science,University of South China,Hengyang 421001,China)

机构地区：[1]南华大学计算机学院,衡阳421001

出　　处：《现代计算机》2023年第23期42-46,共5页Modern Computer

基　　金：南华大学校级大学生创新创业训练项目(D202305182005285040)。

摘　　要：对生物医学领域中影像图片识别的问题进行了研究,提出了融合CTPN和CRNN模型的影像图片文字识别方法。首先利用CTPN模型检测文字所在位置,再利用卷积循环神经网络进行特征序列提取和识别,并基于BLSTM学习序列特征将得分高的窗口连接作为文本区域,最后得到文本的检测识别结果。研究结果表明,融合后能够高效地定位和识别医学影像中的文字信息,为医学影像图片的处理提供了一种有效的解决方案。The problem of image picture recognition in the biomedical field has been studied,and the method of fusing CTPN and CRNN models for text recognition of image pictures is proposed.Firstly,the CTPN model is used to detect the location of text,then the convolutional recurrent neural network is used for feature sequence extraction and recognition and learning sequence fea-tures based on BLSTM,and the windows with high scores are connected as text regions,and finally the detection and recognition results of text are obtained.The results show that the fusion can efficiently locate and recognize text information in medical images,which provides an effective solution for the processing of medical image pictures.

关 键 词：影像图片 文字检测 卷积循环神经网络 文字识别 

分 类 号：R-05[医药卫生] TP391.41[自动化与计算机技术—计算机应用技术]