基于深度学习的光学遥感机场与飞行器目标识别技术

doi:10.11959/j.issn.2096-0271.2016054

摘要/Abstract

摘要：

机场与飞行器目标识别是遥感数据分析中的典型应用。研究了光学遥感大数据环境下面向机场与飞行器目标识别的深度学习技术。为此，构建了一个面向高分光学遥感图像的机场与飞行器目标秒级识别系统。使用迁移学习的方法在有标签样本稀缺的情况下有效构建深度网络，利用目标先验知识对潜在目标进行高效提取，并提出一种层次式的级联深度网络识别架构，实现“大范围、小目标”的实时识别。实验结果表明，采用相应技术，基于深度学习方法可以在秒级时间得到比传统方法更高的识别精度。

关键词: 光学遥感, 目标识别, 深度学习

Abstract:

Airport and flight recognition are the typical remote sensing applications.For the big optical remote sensing data,deep learning techniques for airport and flight recognition have been studied.To this end,a seconds’ response airport and flight recognition system for optical remote sensing data was built.To obtain effective deep learning model with limited labeled samples,transfer learning approach has been employed.Prior knowledge has also been explored for efficient object proposal.To achieve real-time performance for such recognition with “large region and small targets”,a cascade framework of deep networks has been proposed.The results of experiments show that,by the proposed deep learning approaches,significant improvement on recognition accuracy could be achieved with seconds’ response.

Key words: optical remote sensing, object recognition, deep learning

牛新, 窦勇, 张鹏, 曹玉社. 基于深度学习的光学遥感机场与飞行器目标识别技术[J]. 大数据, 2016, 2(5): 54-67.

Xin NIU, Yong DOU, Peng ZHANG, Yushe CAO. Airport and flight recognition on optical remote sensing data by deep learning[J]. Big Data Research, 2016, 2(5): 54-67.

图/表 14

图1

图2

图3

图4

图5

表1

图6

图7

图8

图9

图10

图11

图12

图13

参考文献 29

[1]	李国庆, 邬延辉 . 大数据时代的对地观测科学研究[J]. 计算机学会通讯, 2013,9(9): 27-31. LI G Q , WU Y H . Research on earth observation in age of big data[J]. Communications of CCF, 2013,9(9): 27-31.
[2]	SUN H , SUN X , WANG H ,et al. Automatic target detection in highresolution remote sensing images using spatial sparse coding bag-of-words model[J]. IEEE Geoscience & Remote Sensing Letters, 2012,9(1): 109-113.
[3]	ZHANG W , SUN X , FU K ,et al. Object detection in high-resolution remote sensing images using rotation invariant parts based model[J]. IEEE Geoscience & Remote Sensing Letters, 2014,11(1): 74-78.
[4]	BAI X , ZHANG H , ZHOU J . VHR object detection based on structural feature extraction and query expansion[J]. IEEE Transactions on Geoscience & Remote Sensing, 2014,52(10): 6508-6520.
[5]	CHEN X , XIANG S , LIU C L ,et al. Aircraft detection by deep belief nets[C]// Iapr Asian Conference on Pattern Recognition,November 5-8,2013,Okinawa,Japan. New Jersey: IEEE Press, 2013：54-58.
[6]	HAN J , ZHANG D , CHENG G ,et al. Object detection in optical remote sensing images based on weakly supervised learning and high-level feature learning[J]. IEEE Transactions on Geoscience &Remote Sensing, 2015,53(6): 3325-3337.
[7]	HU X , SHEN J , SHAN J ,et al. Local edge distributions for detection of salient structure textures and objects[J]. IEEE Geoscience & Remote Sensing Letters, 2013,10(3): 466-470.
[8]	DIAO W , SUN X , ZHENG X ,et al. Efficient saliency-based object detection in remote sensing images using deep belief networks[J]. IEEE Geoscience & Remote Sensing Letters, 2016,13(2): 137-141.
[9]	ZHOU P , CHENG G , LIU Z ,et al. Weakly supervised target detection in remote sensing images based on transferred deep features and negative bootstrapping[J]. Multidimensional Systems & Signal Processing, 2016,27(4): 925-944.
[10]	SUN X , WANG H , FU K . Automatic detection of geospatial objects using taxonomic semantics[J]. IEEE Geoscience &Remote Sensing Letters, 2010,7(1): 23-27.
[11]	INGLADA J , MICHEL J . Qualitative spatial reasoning for high-resolution remote sensing image analysis[J]. IEEE Transactions on Geoscience & Remote Sensing, 2009,47(2): 599-612.
[12]	SUN H , SUN X , WANG H ,et al. Automatic target detection in highresolution remote sensing images using a contour-based spatial model[J]. IEEE Geoscience & Remote Sensing Letters, 2012,9(5): 886-890.
[13]	LECUN Y , BOSER B , DENKER J S ,et al. Backprop-agation applied to handwritten zip code recognition[J]. Neural Computation, 1989,1(4): 541-551.
[14]	GIRSHICK R , DONAHUE J , DARRELL T ,et al. Rich feature hierarchies for accurate object detection and semantic segmentation[C]// 2014 IEEE Conference on Computer Vision and Pattern Recognition (CVPR),June 23-28,2014,Columbus,OH,USA. New Jersey: IEEE Press, 2014：580-587.
[15]	HE K , ZHANG X , REN S ,et al. Spatial pyramid pooling in deep convolutional networks for visual recognition[J]. IEEE Transactions on Pattern Analysis & Machine Intelligence, 2014,37(9): 1904-1916.
[16]	GIRSHICK R , . Fast R-CNN[C]// The IEEE International Conference on Computer Vision,December 7-13,2015,Chile. New Jersey: IEEE Press, 2015：1440-1448.
[17]	REN S , HE K , GIRSHICK R ,et al. Faster R-CNN:towards real-time object detection with region proposal networks[J]. Computer Science, 2015(99):1.
[18]	UIJLINGS J R R , SANDE K , GEVERS T ,et al. Selective search for object recognition[J]. International Journal of Computer Vision, 2013,104(2): 154-171.
[19]	ZITNICK C L , DOLL R P . Edge boxes:locating object proposals from edges[M]// Computer Vision-ECCV 2014. Switzerland: Springer International Publishing, 2014:391-405.
[20]	CHENG M M , ZHANG Z , LIN W Y ,et al. BING:binarized normed gradients for objectness estimation at 300fps[C]// IEEE Conference on Computer Vision and Pattern Recognition,June 23-28,2014,Columbus,OH,USA. New Jersey: IEEE Press, 2014：3286-3293.
[21]	REDMON J , DIVVALA S , GIRSHICK R ,et al. You only look once:unified,realtime object detection[J]. Computer Science, 2015:arXiv:1506.02640.
[22]	LIU W , ANGUELOV D , ERHAN D ,et al. SSD:single shot multibox detector[J]. Computer Science, 2015:arXiv:1512.02325.
[23]	HINTON G E . Deep belief networks[J]. Scholarpedia, 2009,4(6): 804-786.
[24]	YU Y , GUAN H , JI Z . Rotation-invariant object detection in high-resolution satellite imagery using superpixel-based deep hough forests[J]. IEEE Geoscience &Remote Sensing Letters, 2015,12(11):1.
[25]	KRIZHEVSKY A , SUTSKEVER I , HINTON G E . Imagenet classification with deep convolutional neural networks[J]. Advances in Neural Information Processing Systems, 2012,25(2): 1-9.
[26]	MCCALLUM . BOW:a toolkit for statistical language modeling,text retrieval,classification and clustering[J]. Machine Learning, 1996(1): 405-408.
[27]	OGAWA S , HIRAO K , HIRAO T , et al. Line segment detector:US,US 20020154121 A1[P]. 2002,
[28]	ZHU D , WANG B , ZHANG L . Airport target detection in remote sensing images:a new method based on two-way saliency[J]. IEEE Geoscience & Remote Sensing Letters, 2015,12(5): 1096-1100.
[29]	QU Y , LI C , ZHENG N . Airport detection base on support vector machine from a single image[C]// Fifth International Conference on Information,Communications and Signal Processing,December 6-9,2005,Bangkok,Thailand. New Jersey: IEEE Press, 2005：546-549.

候选方法	准确率	虚警率	平均耗时/s
基于显著性区域候选方法^[28]+CNN	68.5%	5%	3.85
基于显著性区域候选方法^[28]+SIFT	52%	8.5%	8.57
笔者提出的区域候选方法+ 8 纹理特征SVM^[29]	62%	7.5%	68.64
笔者提出的区域候选方法+ SIFT	63%	15%	64.09
笔者提出的区域候选方法+ CNN	87%	12%	21.3