帮助 关于我们

返回检索结果

Dense RFB和LSTM遥感图像舰船目标检测
Ship detection in remote sensing image based on dense RFB and LSTM

查看参考文献29篇

张涛 1   杨小冈 1   卢孝强 2   卢瑞涛 1   张胜修 1  
文摘 针对当前遥感图像舰船目标检测精度不佳问题,本文构建舰船目标数据集STAR,提出基于Dense RFB和LSTM多尺度舰船目标检测算法。该算法首先在SSD网络基础上设计了浅层特征增强模块,基于人眼视点图采用Dense RFB特征复用和膨胀卷积增大感受野的尺度和种类,增强浅层网络对细节特征的提取能力;其次设计了深层多尺度特征金字塔融合模块,采用FPN和LSTM思想,基于反卷积和残差网络对深层不同尺度特征进行融合,增强网络结构非线性和特征层的表征能力;最后加入聚焦分类损失函数进行联合训练,有效避免了正负样本失衡问题。在遥感图像舰船数据集上实验,本文所提舰船目标检测算法精度均值达到81.98%,检测速度达到29.6帧/s。此外,遥感图像中成像模糊、被遮挡、部分被裁剪等舰船目标的检测效果也优于原有经典算法,实验结果表明该算法对遥感图像舰船目标检测的泛化能力较强,有效地提高了遥感图像舰船目标检测的精度。
其他语种文摘 Ship detection plays a crucial role in various applications and has drawn increasing attention in recent years. Deep learning methods based on CNNs, particularly SSD, have greatly improved detection performance due to their highly efficient feature extraction capability. However, SSD still has two problems. For instance, the detection network of arbitrarily arranged ship targets lacks a connection between high and low-level features and ignores contextual semantic information. Another problem is that natural factors such as light and clouds affect remote sensing images, thus ship detection may cause an imbalance of positive and negative samples. Aiming at solving the above issues, this paper proposes to achieve ship detection in remote sensing images by using a method based on Dense RFB and LSTM. This proposed method includes three elements. First, to enhance the detail feature extraction capability, this proposed method introduces a shallow feature enhancement module. This module draws on the idea of the human viewpoint, which uses Dense RFB feature reuse and expansion convolution to increase the receptive field. Second, to effectively extract deep semantic information and enhance the expressive ability of the network feature layer, a deep multi-scale feature pyramid fusion module (MFPF)is designed, as this proposed method draws on FPN and LSTM deconvolution and residual structure fuse deep multi-scale features. Finally, to solve the imbalance of positive and negative samples, the focal classification loss function is added, improving the accuracy of ship detection during training. The experiments were carried out on an optical remote sensing image dataset, in which only the ship dataset was used for training, validation, and testing. Results indicate that the proposed algorithm achieved an Average Precision (AP)of 81.98% and the detection speed reached 29.6 fps for ship targets, in which most ships were detected successfully. Moreover, for blurred, occluded, and partially-cropped ship targets, the algorithm's detection effect is better than the traditional algorithm. Qualitative and quantitative results indicate that the generalization capability of the proposed method enhances ship detection. From this paper, we can draw three conclusions: (1)The proposed method can improve the extraction of detailed features and increase the receptive fields. (2)The focal loss function method shows good generalization capability. (3)The rotating box detection method is suitable for multi-scale and densely-arranged remote sensing images.
来源 遥感学报 ,2022,26(9):1859-1871 【核心库】
DOI 10.11834/jrs.20211042
关键词 舰船目标检测 ; Dense RFB ; 特征金字塔 ; LSTM ; 多尺度特征
地址

1. 火箭军工程大学导弹工程学院, 西安, 710025  

2. 中国科学院西安光学精密机械研究所, 西安, 710068

语种 中文
文献类型 研究性论文
ISSN 1007-4619
学科 社会科学总论;自动化技术、计算机技术;水路运输
基金 国家自然科学基金 ;  中国航空科学基金 ;  陕西省自然科学基金
文献收藏号 CSCD:7320972

参考文献 共 29 共2页

1.  Cao G M. Feature-fused SSD: fast detection for small objects. Proceedings of SPIE 10615, Ninth International Conference on Graphic and Image Processing,2018:106151E CSCD被引 2 次    
2.  董超. 可见光遥感图像海面舰船目标检测技术研究,2020 CSCD被引 2 次    
3.  郭威. 基于深度学习的光学遥感图像自动舰船检测,2019:15-20 CSCD被引 1 次    
4.  He K M. Deep residual learning for image recognition. Proceedings of the 2016 IEEE Conference on Computer Vision and Pattern Recognition,2016:770-778 CSCD被引 305 次    
5.  Krizhevsky A. ImageNet classification with deep convolutional neural networks. Communications of the ACM,2017,60(6):84-90 CSCD被引 3320 次    
6.  李晖晖. 基于CReLU和FPN改进的SSD舰船目标检测. 仪器仪表学报,2020,41(4):183-190 CSCD被引 24 次    
7.  Lin T Y. Feature pyramid networks for object detection. 2017 IEEE Conference on Computer Vision and Pattern Recognition,2017:936-944 CSCD被引 130 次    
8.  Lin T Y. Focal loss for dense object detection. IEEE Transactions on Pattern Analysis and Machine Intelligence,2020,42(2):318-327 CSCD被引 566 次    
9.  Liu S T. Receptive field block net for accurate and fast object detection. 15th European Conference on Computer Vision,2018:385-400 CSCD被引 1 次    
10.  Liu W. SSD: single shot MultiBox detector. 14th European Conference on Computer Vision,2016:21-27 CSCD被引 101 次    
11.  罗会兰. 基于深度学习的目标检测研究综述. 电子学报,2020,48(6):1230-1239 CSCD被引 65 次    
12.  马健. 基于特征融合SSD的遥感图像舰船目标检测. 计算机应用,2019,39(S2):253-256 CSCD被引 5 次    
13.  Redmon J. You only look once: unified, real-time object detection. 2016 IEEE Conference on Computer Vision and Pattern Recognition,2016:779-788 CSCD被引 219 次    
14.  Ren S Q. Faster R-CNN: towards real-time object detection with region proposal networks. IEEE Transactions on Pattern Analysis and Machine Intelligence,2017,39(6):1137-1149 CSCD被引 4540 次    
15.  史文旭. 基于特征融合的遥感图像舰船目标检测方法. 光子学报,2020,49(7):57-67 CSCD被引 5 次    
16.  史文旭. 特征增强SSD算法及其在遥感目标检测中的应用. 光子学报,2020,49(1):154-163 CSCD被引 11 次    
17.  王伦文. 光学遥感图像目标检测方法. 系统工程与电子技术,2019,41(10):2163-2169 CSCD被引 3 次    
18.  王玺坤. 基于改进型YOLO算法的遥感图像舰船检测. 北京航空航天大学学报,2020,46(6):1184-1191 CSCD被引 17 次    
19.  谢学立. 基于动态感受野的航拍图像目标检测算法. 光学学报,2020,40(4):107-119 CSCD被引 2 次    
20.  于野. A-FPN算法及其在遥感图像船舶检测中的应用. 遥感学报,2020,24(2):107-115 CSCD被引 14 次    
引证文献 3 篇

1 赵其昌 光学遥感图像舰船目标检测与识别方法研究进展 航空学报,2024,45(8):029025
CSCD被引 0 次

2 郭柏麟 基于脉冲神经网络微调方法的遥感图像目标检测 遥感学报,2024,28(7):1702-1712
CSCD被引 0 次

显示所有3篇文献

论文科学数据集
PlumX Metrics
相关文献

 作者相关
 关键词相关
 参考文献相关

版权所有 ©2008 中国科学院文献情报中心 制作维护：中国科学院文献情报中心
地址：北京中关村北四环西路33号 邮政编码：100190 联系电话：(010)82627496 E-mail:cscd@mail.las.ac.cn 京ICP备05002861号-4 | 京公网安备11010802043238号