一种基于自动特征学习的陨石坑区域检测算法

陆婷婷; 张尧; 阎岩; 杨利民; 杨卫东

doi:10.13700/j.bh.1001-5965.2020.0109

一种基于自动特征学习的陨石坑区域检测算法

doi: 10.13700/j.bh.1001-5965.2020.0109

1.
中国运载火箭技术研究院研究发展部, 北京 100176
2.
河南工业大学粮食信息处理与控制教育部重点实验室, 郑州 450001

详细信息

作者简介:
陆婷婷女, 博士, 工程师。主要研究方向: 视觉导航、图像处理、人工智能、指挥控制

通讯作者:
陆婷婷, E-mail:tingtingspring@163.com

中图分类号: TP751.1;V476.4
计量
- 文章访问数: 773
- HTML全文浏览量: 190
- PDF下载量: 209
- 被引次数: 0
出版历程
- 收稿日期: 2020-03-22
- 录用日期: 2020-04-17
- 网络出版日期: 2021-05-20

A crater region detection algorithm based on automatic feature learning

1.
Research and Development Department, China Academy of Launch Vehicle Technology, Beijing 100176, China
2.
Key Laboratory of Grain Information Processing and Control, Ministry of Education, Henan University of Technology, Zhengzhou 450001, China

More Information

Corresponding author: LU Tingting, E-mail:tingtingspring@163.com

摘要

摘要:
基于陨石坑的视觉导航技术成为一种新颖的高精度空间探测自主导航方式，如何从导航图像中精确地提取陨石坑区域是实现基于陨石坑视觉导航的首要条件。针对这一问题，根据陨石坑导航图像特点，提出了一种基于自动特征学习的陨石坑区域检测算法。首先，基于最大稳定极值区域检测算法提取陨石坑候选区域；其次，利用卷积神经网络（CNN）自动学习提取候选区域的特征；最后，通过支持向量机（SVM）实现候选区域的精确分类，得到真实的陨石坑区域。大量的仿真实验表明：与传统的基于人工特征的陨石坑区域检测算法相比，提出的基于自动特征学习的陨石坑区域检测算法具有更高的检测精度和更好的鲁棒性，在通用火星表面陨石坑数据集上，所提算法的F₁度量指标较于传统算法高出8%，可以广泛地应用于基于陨石坑的视觉导航算法中的陨石坑区域提取，为基于陨石坑视觉导航算法提供精确的导航路标输入。
- 陨石坑区域 /
- 目标检测 /
- 自动特征学习 /
- 深度学习 /
- 人工智能
Abstract:
The crater-based navigation technology has been become a novel and precise autonomous navigation method in space exploration, and how to extract the crater regions from the crater navigation image is the essential condition of the crater-based navigation method. Accordingly, in this paper, we propose an algorithm for extracting crater regions via automatic feature learning. First, the candidate crater regions were obtained by the maximal stable external region method. Then, the features of these regions were automatically extracted by Convolutional Neural Network (CNN). Finally, the true crater regions were identified from all the candidate regions through Support Vector Machine (SVM) classifier. The experimental results demonstrate that the proposed algorithm can extract crater regions from the navigation image with higher accuracy and robustness than the traditional crater region detection algorithms based on the handcrafted features. The proposed algorithm obtains an F₁ score which is 8% higher than that of the traditional method on the standard Mars surface crater database, and can be applied in the crater detection of the crater-based visual navigation method to provide the precise navigation landmarks.
- crater region /
- object detection /
- automatic feature learning /
- deep learning /
- artificial intelligence

HTML全文

图 1 陨石坑区域检测算法总体框架

Figure 1. Overall framework of crater region detection algorithm

下载: 全尺寸图片幻灯片

图 2 图像陨石坑的光照特征

Figure 2. Optical characteristics of imaged crater

下载: 全尺寸图片幻灯片

图 3 Crater MSER算法的基本流程

Figure 3. Basic flowchart of Crater MSER algorithm

下载: 全尺寸图片幻灯片

图 4 Crater MSER算法基本流程的真实实例

Figure 4. Instances of basic procedure of Crater MSER algorithm

下载: 全尺寸图片幻灯片

图 5 极小稳定区域平均灰度值的一维K均值聚类示意图

Figure 5. Schematic diagram of one-dimensional K-means cluster of average gray values of minimal MSER regions

下载: 全尺寸图片幻灯片

图 6 CraterCNN网络结构及基于SVM的陨石坑候选区域分类

Figure 6. Structure of CraterCNN network and crater candidate region classification based on SVM

下载: 全尺寸图片幻灯片

图 7 陨石坑候选区域提取实验结果

Figure 7. Experimental results of crater candidate region extraction

下载: 全尺寸图片幻灯片

图 8 h0905_0000陨石坑数据集中的正负样本示例

Figure 8. Positive and negative examples in h0905_0000crater dataset

下载: 全尺寸图片幻灯片

图 9 10-Fold训练测试数据集示意图

Figure 9. Schematic diagram of 10-Fold training andtesting dataset

下载: 全尺寸图片幻灯片

图 10 基于t-SNE算法的CraterCNN特征和Haar-like特征可视化结果

Figure 10. Visualized results of CraterCNN features and Haar-like features based on t-SNE algorithm

下载: 全尺寸图片幻灯片

图 11 陨石坑仿真图像构成的陨石坑区域数据集正负样本示例

Figure 11. Positive and negative examples in crater region dataset composed of simulated crater images

下载: 全尺寸图片幻灯片

图 12 基于解析数据的陨石坑仿真图像陨石坑区域检测结果图例

Figure 12. Examples of crater region detection experimental results of crater simulation image based on analytical data

下载: 全尺寸图片幻灯片

图 13 基于真实陨石坑图像陨石坑区域检测结果图例

Figure 13. Examples of crater region detection experimental results based on real crater simulation image

下载: 全尺寸图片幻灯片

表 1 h0905_0000陨石坑数据集分组

Table 1. Groups of h0905_0000 crater dataset

区域	正样本数量	负样本数量
西部区域	1 121	1 385
中部区域	443	738
东部区域	458	765

下载: 导出CSV

表 2 基于交叉校验数据集的软间隔线性SVM分类器惩罚因子选择实验结果

Table 2. Experimental results for choosing the penalty factor of linear SVM classifier with soft interval based on cross check dataset

序号	惩罚因子pe	平均F₁度量值
1	0.000 01	0.832 6
2	0.000 1	0.893 7
3	0.001	0.884 3
4	0.01	0.851 7
5	0.1	0.832 6
6	1	0.832 6
7	10	0.832 6
8	50	0.832 6

下载: 导出CSV

表 3 CraterCNN+SVM与其他陨石坑候选区域分类算法对比实验结果

Table 3. Contrast experimental results of Crater CNN+SVM algorithm and other crater candidate regions classification algorithms

度量指标	特征类型和分类器类型	西部区域	中部区域	东部区域	平均值
查全率P	Haar-like+AdaBoost	0.903 0	0.903 0	0.903 0	0.903 0
	Haar-like+SVM	0.839 8	0.839 8	0.839 8	0.839 8
	CraterCNN+Softmax	0.929 5	0.871 2	0.929 5	0.910 1
	CraterCNN+SVM	0.929 5	0.898 9	0.929 5	0.919 3
	CraterCNN+AdaBoost	0.898 8	0.905 5	0.898 8	0.901 0
查准率R	Haar-like+AdaBoost	0.863 7	0.863 7	0.863 7	0.863 7
	Haar-like+SVM	0.829 1	0.829 1	0.829 1	0.829 1
	CraterCNN+Softmax	0.916 3	0.880 5	0.916 3	0.904 4
	CraterCNN+SVM	0.916 5	0.910 5	0.916 5	0.914 5
	CraterCNN+AdaBoost	0.904 2	0.871 1	0.904 2	0.893 2
F₁	Haar-like+AdaBoost	0.880 9	0.880 9	0.880 9	0.880 9
	Haar-like+SVM	0.832 6	0.832 6	0.832 6	0.832 6
	CraterCNN+Softmax	0.922 0	0.874 5	0.922 0	0.906 2
	CraterCNN+SVM	0.922 6	0.903 9	0.922 6	0.916 4
	CraterCNN+AdaBoost	0.900 4	0.886 7	0.900 4	0.895 8

下载: 导出CSV

表 4 不同特征提取及不同分类算法的训练和测试在10-Fold数据集上的平均时间

Table 4. Mean time of different feature extraction algorithms and different classification algorithms on 10-Fold training dataset and testing dataset

测试对象	平均时间值/s			平均值/s
测试对象	西部区域	中部区域	东部区域	平均值/s
CraterCNN特征提取	1.54	0.74	0.70	0.99
Haar-like特征提取	4.13	2.08	2.33	2.85
CraterCNN+SVM训练	0.34	0.10	0.08	0.17
CraterCNN+Adaboost训练	796.1	191.08	195.83	394.34
CraterCNN+SVM测试	0.01	0.01	0.01	0.01
CraterCNN+Adaboost测试	0.02	0.01	0.01	0.01

下载: 导出CSV

表 5 基于解析数据的陨石坑仿真图像陨石坑候选区域检测结果统计

Table 5. Candidate crater region detection experimental results of crater simulation image based on analytical data

算法	度量指标	仿真图像1(含9个候选区域)	仿真图像2(含11个候选区域)	仿真图像3(含9个候选区域)	仿真图像4(含11个候选区域)
CraterCNN+SV	检测个数	9	11	9	11
MHaar-like+SVM	检测个数	7	10	8	9
本文算法	运行时间/s	35.26	29.17	31.18	33.76
CraterCNN+SV		0.185 3	0.198 6	0.172 8	0.198 5
MHaar-like+SVM		0.240 7	0.264 9	0.231 8	0.253 9

下载: 导出CSV

表 6 基于真实陨石坑图像的陨石坑候选区域检测结果统计

Table 6. Candidate crater region detection experimental results based on real crater image

算法	检测个数
算法	真实陨石坑图像1(含6个候选区域)	真实陨石坑图像2(含7个候选区域)	真实陨石坑图像3(含8个候选区域)	真实陨石坑图像4(含9个候选区域)
CraterCNN+SVM	6	7	8	9
Haar-like+SVM	7	6	7	8

下载: 导出CSV

参考文献(47)

[1]	LU T, HU W, LIU C, et al. Relative pose estimation of a lander using crater detection and matching[J]. Optical Engineering, 2016, 55(2): 1-24. http://adsabs.harvard.edu/abs/2016OptEn..55b3102L
[2]	YU M, CUI H, TIAN Y. A new approach based on crater detection and matching for visual navigation in planetary landing[J]. Advances in Space Research, 2014, 53(12): 1810-1821. doi: 10.1016/j.asr.2013.04.011
[3]	TIAN Y, YU M, YAO M B, et al. Crater edge-based flexible autonomous navigation for planetary landing[J]. The Journal of Navigation, 2019, 72(3): 649-668. doi: 10.1017/S0373463318000966
[4]	YU M, LI S, WANG S, et al. Single crater-aided inertial navigation for autonomous asteroid landing[J]. Advances in Space Research, 2019, 63(2): 1085-1099. doi: 10.1016/j.asr.2018.09.035
[5]	MAASS B, WOICKE S, OLIVEIRA W M, et al. Crater navigation system for autonomous precision landing on the moon[J]. Journal of Guidance, Control, and Dynamics, 2020, 43(11): 1-18. http://www.researchgate.net/publication/340791531_Crater_Navigation_System_for_Autonomous_Precision_Landing_on_the_Moon
[6]	VIJAYAN S, VANI K, SANJEEVI S. Crater detection, classification and contextual information extraction in lunar images using a novel algorithm[J]. Icarus, 2013, 226(1): 798-815. doi: 10.1016/j.icarus.2013.06.028
[7]	JAIN A, SURESH K, THAKORE D, et al. Automatic crater detection on lunar surface[J]. International Journal of Innovative Research in Science Engineering & Technology, 2013, 2(5): 1448-1453. http://www.researchgate.net/publication/288900777_Automatic_crater_detection_on_Lunar_surface
[8]	BANDEIRA L, SARAIVA J, PINA P. Impact crater recognition on mars based on a probability volume created by template matching[J]. IEEE Transactions on Geoscience and Remote Sensing, 2007, 45(12): 4008-4015. doi: 10.1109/TGRS.2007.904948
[9]	LI K, MU L, LIU J, et al.Impact crater detection based on regional segmentation using Chang'E-1 CCD data[C]//Proceedings of International Congress on Image and Signal Processing.Piscataway: IEEE Press, 2011: 1911-1915.
[10]	SAWABE Y, MATSUNAGA T, ROKUGAWA S. Automated detection and classification of lunar craters using multiple approaches[J]. Advances in Space Research, 2006, 37(1): 21-27. doi: 10.1016/j.asr.2005.08.022
[11]	WEISMULLER T, CABALLERO D, LEINZ M.Technology for autonomous optical planetary navigation and precision landing[C]//Proceedings of AIAA SPACE 2007 Conference & Exposition.Reston: AIAA, 2007: 1-26.
[12]	SADHUKHAN P, PALIT S.Fast autonomous crater detection by image analysis-for unmanned landing on unknown terrain[C]//Proceedings of Image and Signal Processing.Berlin: Springer, 2016: 293-303.
[13]	LEROY B, MEDINONI G, JOHNSON E, et al. Crater detection for autonomous landing on asteroids[J]. Image & Vision Computing, 2001, 19(11): 787-792.
[14]	CHENG Y, MILLER J K. Autonomous landmark based spacecraft navigation system[J]. Advances in the Astronautical Sciences, 2003, 114(3): 1769-1783. http://www.mendeley.com/research/autonomous-landmark-based-spacecraft-navigation-system/
[15]	CHENG Y, JOHNSON A E, MATTHIES L H, et al.Optical landmark detection for spacecraft navigation[C]//Proceedings of AAS/AIAA Space Flight Mechanics Meeting.Reston: AIAA, 2003: 1-19.
[16]	MAMMARELLA M, RODRIGALVAREZ M A, PIZZICHINI A, et al. Advances in aerospace guidance, navigation and control[M]. Berlin: Springer, 2011: 419-430.
[17]	HE J, CUI H, FENG J.Edge information based crater detection and matching for lunar exploration[C]//Proceedings of International Conference on Intelligent Control and Information Processing.Piscataway: IEEE Press, 2010: 302-307.
[18]	FENG J, CUI P, CUI H.Autonomous hazard detection and landing point selecting for planetary landing[C]//Proceedings of International Symposium on Systems and Control in Aeronautics and Astronautics.Piscataway: IEEE Press, 2010: 1292-1296.
[19]	DING M, CAO Y, WU Q. Method of passive image based crater autonomous detection[J]. Chinese Journal of Aeronautics, 2009, 22(3): 301-306. doi: 10.1016/S1000-9361(08)60103-X
[20]	DING M, CAO Y F, WU Q X.Autonomous craters detection from planetary image[C]//Proceedings of International Conference on Innovative Computing Information and Control.Piscataway: IEEE Press, 2008: 443-448.
[21]	DING W, STEPINSKI T F, BANDEIRA L, et al.Automatic detection of craters in planetary images: An embedded framework using feature selection and boosting[C]//Proceedings of ACM Conference on Information and Knowledge Management.New York: ACM Press, 2010: 749-758.
[22]	DING W, STEPINSKI T F, MU Y, et al. Subkilometer crater discovery with boosting and transfer learning[J]. ACM Transactions on Intelligent Systems & Technology, 2011, 2(4): 1-22.
[23]	MARTINS R, PINA P, MARQUES J S, et al. Crater detection by a boosting approach[J]. IEEE Geoscience & Remote Sensing Letters, 2009, 6(1): 127-131. http://ieeexplore.ieee.org/document/4717240
[24]	BANDIEIRA L, DING W, STEPINSKI T F. Detection of sub-kilometer craters in high resolution planetary images using shape and texture features[J]. Advances in Space Research, 2012, 49(1): 64-74. doi: 10.1016/j.asr.2011.08.021
[25]	JIN S, ZHANG T. Automatic detection of impact craters on Mars using a modified adaboosting method[J]. Planetary and Space Science, 2014, 99(1): 112-117. http://www.sciencedirect.com/science/article/pii/S0032063314001196
[26]	COHEN J P, DING W. Crater detection via genetic search methods to reduce image features[J]. Advances in Space Research, 2014, 53(12): 1768-1782. doi: 10.1016/j.asr.2013.05.010
[27]	TAKEDA Y, AOYAMA N, TANAAMI T, et al. Study on crater detection using FLDA[J]. World Academy of Science Engineering & Technology, 2012, 6(11): 1373-1377. http://www.waset.org/Publication/study-on-crater-detection-using-flda/15708
[28]	LENA D, TED J S, JONATHAN P H.Deep learning crater detection for lunar terrain relative navigation[C]//Proceedings of AIAA Scitech 2020 Forum.Reston: AIAA, 2020: 1-12.
[29]	EMAMI E, AHMAD T, BEBIS G, et al. Crater detection using unsupervised algorithms and convolutional neural networks[J]. IEEE Transactions on Geoscience and Remote Sensing, 2019, 57(8): 5373-5383. doi: 10.1109/TGRS.2019.2899122
[30]	LEE C. Automated crater detection on Mars using deep learning[J]. Planetary and Space Science, 2019, 170: 16-28. doi: 10.1016/j.pss.2019.03.008
[31]	NISTER D, STEWENIUS H.Linear time maximally stable extremal regions[C]//Proceeding of 10th European Conference on Computer Vision.Berlin: Springer, 2008: 183-196.
[32]	ROERDINK J B T M, MEIJSTER A. The watershed transform: Definitions, algorithms and parallelization strategies[J]. Fundamenta Informaticae, 2000, 41(1): 187-228. http://portal.acm.org/citation.cfm?id=341161
[33]	SIMONYAN K, ZISSERMAN A.Very deep convolutional networks for large-scale image recognition[C]//Procceeding of ICLR, 2015: 1-14.
[34]	CHERON G, LAPTEV I, SCHMID C.P-CNN: Pose-based CNN features for action recognition[C]//Proceedings of IEEE International Conference on Computer Vision.Piscataway: IEEE Press, 2015: 3218-3226.
[35]	KOBCHAISAWAT T, CHALIDABHONGSE T H.Thai text localization in natural scene images using convolutional neural network[C]//Proceedings of Signal and Information Processing Association Annual Summit and Conference (APSIPA).Piscataway: IEEE Press, 2015: 1-7.
[36]	RANJAN R, PATEL V M, CHELLAPPA R.A deep pyramid deformable part model for face detection[C]//Proceedings of IEEE International Conference on Biometrics Theory, Applications and Systems.Piscataway: IEEE Press, 2015: 1-8.
[37]	SHARIF R A, AZIZPOUR H, SULLIVAN J, et al.CNN features off-the-shelf: An astounding baseline for recognition[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition Workshops.Piscataway: IEEE Press, 2014: 806-813.
[38]	VIOLA P A, JONES M J.Detecting pedestrians using patterns of motion and appearance in videos: US7212651B2[P].2007-05-01.
[39]	DALAL N, TRIGGS B.Histograms of oriented gradients for human detection[C]//Proceedings of Computer Vision and Pattern Recognition.Piscataway: IEEE Press, 2005: 886-893.
[40]	PIETIKINEN M, HADID A, ZHAO G, et al. Computer vision using local binary patterns[M]. Berlin: Springer, 2011: 3-10.
[41]	SZARVAS M, YOSHIZAWA A, YAMAMOTO M, et al.Pedestrian detection with convolutional neural networks[C]//Proceedings of Intelligent Vehicles Symposium.Piscataway: IEEE Press, 2005: 224-229.
[42]	NIU X X, SUEN C Y. A novel hybrid CNN-SVM classifier for recognizing handwritten digits[J]. Pattern Recognition, 2012, 45(4): 1318-1325. doi: 10.1016/j.patcog.2011.09.021
[43]	ZHANGL L, SHI Z, WU J. A hierarchical oil tank detector with deep surrounding features for high-resolution optical satellite imagery[J]. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2015, 8(10): 1-15. doi: 10.1109/JSTARS.2016.2514638
[44]	Math Works.Pretrained convolutional neural networks[EB/OL].[2020-02-24].https://cn.mathworks.com/help/nnet/ug/pretrained-convolutional-neural-networks.html.
[45]	Knowledge Discovery Lab.Crater datbase[EB/OL].[2020-02-15].http://kdl.cs.umb.edu/w/datasets/craters/.
[46]	LU T, HU W, JIANG Z.An effective algorithm for generation of crater gray image[C]//Proceedings of IEEE International Conference on Computational Intelligence and Virtual Environments for Measurement Systems and Applications.Piscataway: IEEE Press, 2015: 1-6.
[47]	陆婷婷, 李潇, 张尧, 等. 基于三维点云模型的空间目标光学图像生成技术[J]. 北京航空航天大学学报, 2020, 46(2): 274-286. doi: 10.13700/j.bh.1001-5965.2019.0189 LU T T, LI X, ZHANG Y, et al. A technology for generation of space object optical image based on 3D point cloud model[J]. Journal of Beijing University of Aeronautics and Astronautics, 2020, 46(2): 274-286(in Chinese). doi: 10.13700/j.bh.1001-5965.2019.0189