|本期目录/Table of Contents|

[1]宋彭彭,曾祥进*,郑安义,等.基于DenseNet的自然场景文本检测[J].武汉工程大学学报,2022,44(03):309-314.[doi:10.19843/j.cnki.CN42-1779/TQ.202106001]
 SONG Pengpeng,ZENG Xiangjin*,ZHENG Anyi,et al.Natural Scene Text Detection Based on DenseNet[J].Journal of Wuhan Institute of Technology,2022,44(03):309-314.[doi:10.19843/j.cnki.CN42-1779/TQ.202106001]
点击复制

基于DenseNet的自然场景文本检测(/HTML)
分享到:

《武汉工程大学学报》[ISSN:1674-2869/CN:42-1779/TQ]

卷:
44
期数:
2022年03期
页码:
309-314
栏目:
机电与信息工程
出版日期:
2022-06-30

文章信息/Info

Title:
Natural Scene Text Detection Based on DenseNet
文章编号:
1674 - 2869(2022)03 - 0309 - 06
作者:
宋彭彭曾祥进*郑安义米 勇
武汉工程大学计算机科学与工程学院,湖北 武汉 430205
Author(s):
SONG Pengpeng ZENG Xiangjin* ZHENG Anyi MI Yong
School of Computer Science and Engineering, Wuhan Institute of Technology, Wuhan 430205, China
关键词:
自然场景文本检测DenseNet协调注意力特征融合
Keywords:
natural scene text detection DenseNet coordinate attention feature fusion
分类号:
TP391
DOI:
10.19843/j.cnki.CN42-1779/TQ.202106001
文献标志码:
A
摘要:
针对自然场景中由文本背景复杂、文字大小不同而引起的文本检测准确率不高的问题,提出了一种基于DenseNet改进的文本检测方法。首先使用DenseNet网络提取更深层次的文本特征,通过引入协调注意力,将位置信息嵌入通道注意力中获取大区域特征;其次对DenseNet网络使用特征融合技术,使改进后的网络能够提取文本信息更丰富的特征,降低了漏检和误检文本的概率。结果表明:该模型在数据集ICDAR2011和ICDAR2013中的准确率分别达到了0.88和0.89,证实了该改进方法的有效性。
Abstract:
For the problems of low accuracy in text detection caused by complex text background and different text sizes in natural scenes, we proposed an improved text detection method based on DenseNet. Firstly ,we used DenseNet network to extract deeper text features, in which the Coordinate attention was introduced to enhance the text features and the location information was embedded into the channel attention to obtain the large area features. Secondly, we employed feature fusion technology for DenseNet network, in which the improved network can extract richer features of text information and reduce the probability of missing and false detection of text. The experiments show that the accuracy of the model reaches 0.88 and 0.89 respectively under the data sets ICDAR2011 and ICDAR2013, which confirms the effectiveness of the improved method.

参考文献/References:

[1] 吴泽俊,赵彤洲.基于区域显著性与稳定性的小目标检测方法[J].武汉工程大学学报,2020,42(3):332-337.

[2] 王润民,桑农,丁丁,等.自然场景图像中的文本检测综述[J].自动化学报,2018,44(12):2113-2141.
[3] 白翔,杨明锟,石葆光,等.基于深度学习的场景文字检测与识别[J].中国科学:信息科学,2018,48(5):531-544.
[4] 郭闯,邱晓晖.基于BLSTM网络的改进EAST文本检测算法[J].计算机技术与发展,2020,30(7):21-24.
[5] SHI B G, BAI X, BELONGIE S. Detecting oriented text in natural images by linking segments[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Honolulu: IEEE Computer Society, 2017: 2550-2558.
[6] 许光宇,尹孟园.基于空间通道注意力的改进SSD目标检测算法[J].光电子·激光,2021,32(9):970-978.
[7] 郝聚涛,段静文,陈超,等.一种基于CTPN网络的文档图像标题检测算法[J].电子技术与软件工程,2021(5):175-176.
[8] TANG J, YANG Z B, WANG Y P, et al. Seglink++: detecting dense and arbitrary-shaped scene text by instance-aware component grouping[J]. Pattern recognition, 2019, 96: 106954.
[9] LONG S B, RUAN J Q, ZHANG W J, et al. Textsnake: a flexible representation for detecting text of arbitrary shapes[C]//Proceedings of the European conference on computer vision (ECCV). Munich: Springer, 2018: 20-36.
[10] 刘会江,曾浩,陈阳.基于DenseNet自然场景文本检测[J].计算机工程与设计,2020,41(8):2201-2206.
[11] HOU Q B, ZHOU D Q, FENG J S. Coordinate attention for efficient mobile network design[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Nashville: IEEE, 2021: 13713-13722.
[12] 李克文,李新宇.基于SENet改进的Faster R-CNN行人检测模型[J].计算机系统应用,2020,29(4):266-271.
[13] 杨锶齐,易尧华,汤梓伟,等.嵌入注意力机制的自然场景文本检测方法[J].计算机工程与应用,2021,57(24):185-191.
[14] LIU J J, HOU Q B, CHENG M M, et al. Improving convolutional networks with self-calibrated convolutions[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. Seattle: IEEE, 2020: 10096-10105.
[15] LOULOUDIS G, STAMATOPOULOS N, GATOS B. ICDAR 2011 writer identification contest[C]//2011 International Conference on Document Analysis and Recognition. Beijing: IEEE, 2011: 1475-1479.
[16] KARATZAS D, SHAFAIT F, UCHIDA S, et al. ICDAR 2013 robust reading competition[C]//2013 12th International Conference on Document Analysis and Recognition. Washington: IEEE, 2013: 1484-1493.
[17] LIAO M H, ZHU Z, SHI B G, et al. Rotation-sensitive regression for oriented scene text detection[C]//Proceedings of the IEEE conference on computer vision and pattern recognition. Salt Lake City: IEEE, 2018: 5909-5918.
[18] 杨剑锋,王润民,何璇,等.基于FCN的多方向自然场景文字检测方法[J].计算机工程与应用,2020,56(2):164-170.
[19] 易尧华,何婧婧,卢利琼,等.顾及目标关联的自然场景文本检测[J].中国图象图形学报,2020,25(1):126-135.
[20] CHENG C K, CHAN C S, LIU C L. Total-text: toward orientation robustness in scene text detection[J]. International Journal on Document Analysis and Recognition (IJDAR), 2020, 23(1): 31-52.

相似文献/References:

备注/Memo

备注/Memo:
收稿日期:2021-06-01
基金项目:国家自然科学基金(61502354);湖北省三峡实验室创新基金(SC215001)
作者简介:宋彭彭,硕士研究生。E-mail:1170552818@qq.com
*通讯作者:曾祥进,博士,副教授。E-mail:xjzeng21@163.com
引文格式:宋彭彭,曾祥进,郑安义,等. 基于DenseNet的自然场景文本检测[J]. 武汉工程大学学报,2022,44(3):309-314.

更新日期/Last Update: 2022-06-29