[1]岳霄,景诗云,史伟.基于改进DenseNet的西夏文识别研究[J].计算机技术与发展,2024,34(10):46-52.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0181]
　YUE Xiao,JING Shi-yun,SHI Wei.Study on Recognition of Xixia Text Based on Improved DenseNet[J].,2024,34(10):46-52.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0181]
点击复制

基于改进DenseNet的西夏文识别研究()

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
34
期数:
2024年10期
页码:
46-52
栏目:
媒体计算
出版日期:
2024-10-10

文章信息/Info

Title:
Study on Recognition of Xixia Text Based on Improved DenseNet
文章编号:
1673-629X(2024)10-0046-07
作者:
岳霄; 景诗云; 史伟
宁夏大学 信息工程学院,宁夏 银川 750021
Author(s):
YUE Xiao; JING Shi-yun; SHI Wei
School of Information Engineering,Ningxia University,Yinchuan 750021,China
关键词:
西夏古籍; 文字识别; 通道重建; 空间重构; 互通道损失
Keywords:
ancient books of Xixia; text recognition; channel reconstruction; spatial reconstruction; mutual-channel loss
分类号:
TP391.1
DOI:
10.20165/j.cnki.ISSN1673-629X.2024.0181
摘要:
由于西夏文字的笔画繁多、结构复杂、相似性高以及西夏古籍存在缺字、狐斑、褪变等问题,目前对其检测识别依然是一项较为困难的研究,现有的识别研究多存在识别精度不够理想、漏检和错检等问题。 因此,在综合分析当前主流研究的基础上,该文提出了一种基于改进 DenseNet(Densely Connected Convolutional Networks)网络模型的西夏文识别方法。该方法通过引入空间通道重构卷积替换原模型中的传统 3×3 卷积,其主要利用通道重建模块和空间重构模块减少了网络训练过程中特征图之间的冗余,提高了网络的特征表示能力;并在损失函数部分采用互通道损失函数代替了交叉熵损失,其在不引入任何外部参数的情况下,进一步减少特征冗余并且提高了网络聚焦于重点识别区域的能力。 通过对比实验的结果表明,在 668 类西夏文识别数据集上,该方法的准确率为 97. 08% ,参数量为 6. 2 MB,相对比于目前主流的方法都有较明显的提升,证明了该方法的有效性。
Abstract:
Due to a large number of strokes,complex structure,high similarity,and the problems of missing characters,foxing,and fading in the ancient books of Xixia,it is still a difficult research to detect and recognize them at present,and the existing recognition studies mostly have problems such as suboptimal recognition accuracy,omission,misdiagnosis. Therefore,we propose an improved DenseNet-based Xixia text recognition method based on a comprehensive analysis of the current mainstream research. The proposed method replaces the traditional 3×3 convolution in the original model by introducing the spatial and channel reconstruction convolution,which mainly utilizes the channel reconstruction module and the spatial reconstruction module to reduce the redundancy between the feature maps in the training process of the network, and improves the feature representation capability of the network. Furthermore, it uses the mutual - channel loss instead of the cross-entropy loss in the loss function part,which further reduces the feature redundancy and improves the ability of the network to focus on the key recognition regions without introducing any external parameters. The results of the comparison experiments show that the accuracy of the proposed method is 97. 08% and the parameters are 6. 2 MB on 668 types of Xixia text recognition datasets,which is a more obvious improvement relative to the current mainstream methods,proving its effectiveness.

相似文献/References:

[1]陈梓洋,王宇飞,钱侃,等. 自然场景下基于区域检测的文字识别算法[J].计算机技术与发展,2015,25(07):230.
　CHEN Zi-yang,WANG Yu-fei,QIAN Kan,et al. Character Recognition Algorithm Based on Region Detection in Natural Scene[J].,2015,25(10):230.
[2]任荣梓,高航. 基于反馈合并的中英文混排版面OCR技术研究[J].计算机技术与发展,2017,27(03):39.
　REN Rong-zi,GAO Hang. Investigation on Layout Analysis Technology of Chinese and English Mixed OCR Based on Feedback Merging[J].,2017,27(10):39.
[3]张婷婷,马明栋,王得玉.OCR 文字识别技术的研究[J].计算机技术与发展,2020,30(04):85.[doi:10. 3969 / j. issn. 1673-629X. 2020. 04. 016]
　ZHANG Ting-ting,MA Ming-dong,WANG De-yu.Research on OCR Technology[J].,2020,30(10):85.[doi:10. 3969 / j. issn. 1673-629X. 2020. 04. 016]
[4]章 安,马明栋.基于 Tesseract 文字识别的预处理研究[J].计算机技术与发展,2021,31(01):73.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 013]
　ZHANG An,MA Ming-dong.Research on Preprocessing Based on Tesseract Text Recognition[J].,2021,31(10):73.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 013]
[5]曾 悦,马明栋.基于 Tesseract_OCR 文字识别的研究[J].计算机技术与发展,2021,31(11):76.[doi:10. 3969 / j. issn. 1673-629X. 2021. 11. 013]
　ZENG Yue,MA Ming-dong.Research on Text Recognition Based on Tesseract_OCR[J].,2021,31(10):76.[doi:10. 3969 / j. issn. 1673-629X. 2021. 11. 013]
[6]蒋子敏,刘宁钟,沈家全.基于轻量级网络的 PCB 芯片文字识别[J].计算机技术与发展,2021,31(12):55.[doi:10. 3969 / j. issn. 1673-629X. 2021. 12. 010]
　JIANG Zi-min,LIU Ning-zhong,SHEN Jia-quan.PCB Chip Text Recognition Based on Lightweight Network[J].,2021,31(10):55.[doi:10. 3969 / j. issn. 1673-629X. 2021. 12. 010]
[7]童 攀,龙炳鑫,拥 措 *.基于注意力机制藏文乌金体古籍文字识别研究[J].计算机技术与发展,2023,33(10):163.[doi:10. 3969 / j. issn. 1673-629X. 2023. 10. 025]
　TONG Pan,LONG Bing-xin,YONG Cuo*.Research on Tibetan Ujin Ancient Book Character Recognition Based on Attention Mechanism[J].,2023,33(10):163.[doi:10. 3969 / j. issn. 1673-629X. 2023. 10. 025]
[8]于海庆,郑廷帅,史伟*.基于改进PSENet的西夏文检测研究[J].计算机技术与发展,2025,(05):16.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0402]
　YU Hai-qing,ZHENG Ting-shuai,SHI Wei*.Research on Xixia Script Detection Based on Improved PSENet[J].,2025,(10):16.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0402]
[9]李子含,屈乐达,刘思源.基于FM-MobileViT网络的拓片甲骨文字识别[J].计算机技术与发展,2025,(05):23.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0020]
　LI Zi-han,QU Le-da,LIU Si-yuan.Rubbing Oracle Bone Character Recognition Based on FM-MobileViT[J].,2025,(10):23.[doi:10.20165/j.cnki.ISSN1673-629X.2025.0020]

更新日期/Last Update: 2024-10-10