首页 > 论文 > 光学学报 > 38卷 > 7期(pp:712006--1)

基于深度卷积神经网络的红外船只目标检测方法

Infrared Ship Target Detection Method Based on Deep Convolution Neural Network

  • 摘要
  • 论文信息
  • 参考文献
  • 被引情况
  • PDF全文
分享:

摘要

针对红外船只图像较模糊导致的识别率低、识别速度慢等问题,提出了一种基于深度卷积神经网络(CNN)的检测算法。首先采用标记分水岭分割算法提取红外船只图像中的连通区域,并对原图相应的目标位置进行标记和归一化处理,提取候选区域。采用改进的AlexNet(一种深度CNN模型)进行船只目标识别,将提取的候选区域送入改进的AlexNet进行特征提取和预测,得到最终检测结果。分水岭方法可大大减少候选区域检测时间,以及减少深度CNN识别时间。利用实验室自制的红外成像系统获取近千张红外船只图像数据,并对其平移缩放形成的数据集进行仿真实验。结果表明,标记分水岭与深度CNN的结合,可有效识别船只目标,所提方法具有良好的性能,能够更加快速准确地识别红外船只目标。

Abstract

Aiming at the problems of low recognition accuracy and slow recognition speed due to the fuzzy image of infrared ship targets, a classification algorithm based on deep convolution neural network (CNN) is proposed. By using the marker-controlled watershed segmentation algorithm, the connected regions in infrared ship image are extracted and the corresponding target positions of the original image are marked and normalized to extract the candidate regions. The improved AlexNet (a deep CNN model) is used for ship targets identification. The extracted candidate regions are sent to the improved AlexNet for feature extraction and prediction to obtain the final detection result. The marker-controlled watershed segmentation method can greatly reduce the number of candidate regions and reduce the classification time of deep CNN. The data of nearly one thousand infrared ship images are obtained by the laboratory-made infrared imaging system, and the simulation experiment on the dataset formed by its translation and scaling is performed. The simulation results show that the combination of the marker-controlled watershed segmentation algorithm and the deep CNN can effectively identify the ship targets. The proposed method has good performance and can identify infrared ship targets more quickly and accurately.

Newport宣传-MKS新实验室计划
补充资料

中图分类号:TP391

DOI:10.3788/aos201838.0712006

所属栏目:仪器,测量与计量

基金项目:天基视频探测技术(2015AAxxx5097)

收稿日期:2018-02-06

修改稿日期:2018-03-09

网络出版日期:--

作者单位    点击查看

王文秀:中国科学院红外探测与成像技术重点实验室, 上海 200083中国科学院上海技术物理研究所, 上海 200083中国科学院大学, 北京 100049
傅雨田:中国科学院红外探测与成像技术重点实验室, 上海 200083中国科学院上海技术物理研究所, 上海 200083
董峰:中国科学院红外探测与成像技术重点实验室, 上海 200083中国科学院上海技术物理研究所, 上海 200083
李锋:中国科学院红外探测与成像技术重点实验室, 上海 200083中国科学院上海技术物理研究所, 上海 200083中国科学院大学, 北京 100049

联系人作者:傅雨田(yutianfu@mail.sitp.ac.cn)

备注:王文秀(1991-),女,博士研究生,主要从事红外成像、图像处理分析等方面的研究。E-mail: wenxiu@mail.ustc.edu.cn

【1】Wang Y Q, Ma L, Tian Y. State-of-the-art of ship detection and recognition in optical remotely sensed imagery[J]. Acta Automatica Sinica, 2011, 37(9): 1029-1039.
王彦情, 马雷, 田原. 光学遥感图像舰船目标检测与识别综述[J]. 自动化学报, 2011, 37(9):1029-1039.

【2】Sun L, Zhou Z Q, Wang B, et al. A novel inshore ship detection via ship head classification and body boundary determination[J]. IEEE Geoscience and Remote Sensing Letters, 2016, 13(12): 1920-1924.

【3】Ren S Q, He K M, Girshick R, et al. Faster R-CNN: towards real-time object detection with region proposal networks[C]. International Conference on Neural Information Processing Systems, 2015: 91-99.

【4】Zhang R Q, Yao J, Zhang K, et al. S-CNN-based ship detection from high-resolution remote sensing images[J]. International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2016, XLI-B7: 423-431.

【5】Zhu S P, Yang L. Stereo matching algorithm with graph cuts based on adaptive watershed[J]. Acta Optica Sinica, 2013, 33(3): 0315004.
祝世平, 杨柳. 基于自适应分水岭的图割的立体匹配算法[J]. 光学学报, 2013, 33(3): 0315004.

【6】Cai Q, Liu Y Q, Cao J, et al. A watershed image segmentation algorithm based on self-adaptive marking and interregional affinity propagation clustering[J]. Acta Electronica Sinica, 2017, 45(8): 1911-1918.
蔡强, 刘亚奇, 曹健, 等. 一种基于自适应标记与区域间近邻传播聚类的分水岭图像分割算法[J]. 电子学报, 2017, 45(8): 1911-1918.

【7】Qu L, Wang K R, Cheng L L, et al. Fast road detection based on RGBD images and convolution neural network[J]. Acta Optica Sinica, 2017, 37(10): 1010003.
曲磊, 王康如,陈利利,等. 基于RGBD图像和卷积神经网络的快速道路检测[J]. 光学学报,2017,37(10):1010003.

【8】Yu W S, Hou Z Q, Wang C Y, et al. Watershed algorithm based on modified filter and marker-extraction[J]. Acta Electronica Sinica, 2011, 39(4): 825-831.
余旺盛, 侯志强, 王朝英, 等. 基于改进滤波和标记提取的分水岭算法[J]. 电子学报, 2011, 39(4): 825-831.

【9】He Z Y, Sun L N, Chen L G. Fast computation of threshold based on Otsu criterion[J]. Acta Electronica Sinica, 2013,41(2): 267-272.
何志勇, 孙立宁, 陈立国. Otsu准则下分割阈值的快速计算[J]. 电子学报, 2013, 41(2): 267-272.

【10】Chang L, Deng X M, Zhou M Q, et al. Convolutional neural networks in image understanding[J]. Acta Automatica Sinica, 2016, 42(9): 1300-1312.
常亮, 邓小明, 周明全, 等. 图像理解中的卷积神经网络[J]. 自动化学报, 2016, 42(9): 1300-1312.

【11】Ye G L, Sun S Y, Gao K J, et al. Nighttime pedestrian detection based on faster region convolution neural network[J]. Laser & Optoelectronics Progress, 2017, 54(8): 081003.
叶国林, 孙韶媛, 高凯珺, 等. 基于加速区域卷积神经网络的夜间行人检测研究[J]. 激光与光电子学进展, 2017, 54(8): 081003.

【12】Xin P, Xu Y L, Tang H, et al. Fast airplane detection based on multi-layer feature fusion of fully convolutional networks[J]. Acta Optica Sinica, 2018, 38(3): 0315003.
辛鹏, 许悦雷, 唐红, 等. 全卷积网络多层特征融合的飞机快速检测[J]. 光学学报, 2018, 38(3): 0315003.

【13】Girshick R, Donahue J, Darrell T, et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]. IEEE Conference on Computer Vision and Pattern Recognition, 2014, 580-587.

【14】Krizhevsky A. One weird trick for parallelizing convolutional neural networks[J]. Neural and Evolutionary Computing, 2014, arXiv: 1404. 5997.

【15】Redmon J, Farhadi A. YOLO9000: better, faster, stronger[J]. IEEE Conference on Computer Vision and Pattern Recognition, 2017: 6517-6525.

【16】Liu F, Shen T S, Ma X X. Convolutional neural network based multi-band ship target recognition with feature fusion[J]. Acta Optica Sinica, 2017, 37(10): 1015002.
刘峰, 沈同圣, 马新星. 特征融合的卷积神经网络多波段舰船目标识别[J]. 光学学报, 2017, 37(10): 1015002.

引用该论文

Wang Wenxiu,Fu Yutian,Dong Feng,Li Feng. Infrared Ship Target Detection Method Based on Deep Convolution Neural Network[J]. Acta Optica Sinica, 2018, 38(7): 0712006

王文秀,傅雨田,董峰,李锋. 基于深度卷积神经网络的红外船只目标检测方法[J]. 光学学报, 2018, 38(7): 0712006

您的浏览器不支持PDF插件,请使用最新的(Chrome/Fire Fox等)浏览器.或者您还可以点击此处下载该论文PDF