بهبود طبقه‌بندی تصاویر ابرطیفی با استفاده از مدل ترکیبی شبکه‌های کپسول و درخت تصمیم تقویتی

نوع مقاله : مقاله پژوهشی

نویسندگان

1 دانشجوی دکتری سنجش از دور، گروه مهندسی نقشه‌برداری و اطلاعات مکانی، دانشگاه صنعتی خواجه نصیرالدین طوسی

2 استادیار،دانشکده مهندسی نقشه برداری واطلاعات مکانی، دانشگاه تهران

3 استاد، گروه مهندسی فتوگرامتری و سنجش ازدور، دانشگاه صنعتی خواجه نصیر طوسی

4 دکتری سنجش از دور، گروه مهندسی فتوگرامتری و سنجش از دور، دانشکده مهندسی نقشه برداری، دانشگاه صنعتی خواجه نصیرالدین طوسی

چکیده

با گسترش دانش سنجش از دور، استفاده از تصاویر هایپراسپکترال روزبه‌روز افزایش و عمومیت می‌یابد. طبقه‌بندی یکی از محبوب‌ترین موضوعات در سنجش از دور ابرطیفی است. طی دو دهة گذشته، روش‌های بسیاری برای مقابله با مشکل طبقه‌بندی داده‌های هایپراسپکترال پیشنهاد شده است. در پژوهش حاضر، ساختاری مبتنی‌بر یادگیری شبکه‌های کپسول برای طبقه‌بندی تصاویر ابرطیفی به‌کار رفته است؛ به‌گونه‌ای که ساختار شبکه بتواند، با استفاده از یک لایة کانولوشنی و یک لایة کپسول، بهترین حالت تولید ویژگی‌ها را داشته باشد و درعین‌حال از بیش‌برازش شبکه روی نمونه‌های آموزشی جلوگیری کند. نتایج به‌دست‌آمده نشان از کیفیت بالای ویژگی‌های تولیدی در ساختار پیشنهادی دارد. درراستای بهبود دقت طبقه‌بندی، رویکرد استخراج ویژگی ازطریق شبکة طراحی‌شده و طبقه‌بندی با استفاده از الگوریتم درخت تقویتی XGBoost، با روش طبقه‌بندی ازطریق شبکة عمیق سراسری مقایسه شد تا، علاوه‌بر بررسی و کیفیت‌سنجی ویژگی‌های عمیق برداری تولیدی به‌روش پیشنهادی در طبقه‌بندی‌کننده‌های گوناگون، میزان توانایی شبکه‌های عمیق سراسری نیز، در کاربرد طبقه‌بندی، بررسی شود. رویکرد کپسول پیشنهادی شامل سه لایة اصلی است: 1)  Prime با کپسول‌هایی به‌اندازة 8 و 32 فیلتر 9×9 و گام حرکتی 2؛ 2) Digitcaps دارای دَه کپسول شانزده‌بعدی؛ 3) لایة تماماً متصل. نتایج بررسی دو رویکرد برای شبکة عمیق و نیز ترکیب شبکه‌های کپسول با الگوریتم درخت تقویتی XGBoost مقایسه شد. رویکردهایی همچون SVM، RF-200، LSTM، GRU، و GRU-Pretanh برای مقایسة رویکرد پیشنهادی براساس پیکربندی‌هایی درنظر گرفته شدند که در تحقیقات به آنها اشاره شده بود. برای ارزیابی مدل پیشنهادی، مجموعه دادة Indian Pines نیز، شامل شانزده کلاس متفاوت، به‌کار رفت. با استفاده از روش پیشنهادی ترکیبی، طبقه‌بندی تصاویر با دقت 99% روی‌داده‌های آموزش و دقت 5/97% روی ‌داده‌های تست انجام می‌شود.

کلیدواژه‌ها


عنوان مقاله [English]

Improving the Classification of Hyperspectral Images Using the Combined Model of CapsNet and the Extreme Gradient Boosting

نویسندگان [English]

  • Pouya Ahmadi 1
  • Tayebe Managhebi 2
  • Hamid Ebadi 3
  • Behnam Asghari 4
1 Ph.D. Student, Faculty of Geomatics Engineering, K.N. Toosi University of Technology, Tehran
2 Assistant Professor, Faculty of Mapping and Spatial Information Engineering, University of Tehran, Tehran, Iran
3 Prof. of Photogrammetry & Remote Sensing Dept,، K.N.Toosi University of Technology،Tehran - Iran
4 Ph. D. in Remote sensing engineering 1 Faculty of Geomatics Engineering , K.N. Toosi University of Technology , Tehran
چکیده [English]

With the development of remote sensing science, the use of hyperspectral images is becoming more widespread. Classification is one of the most popular topics in hyperspectral remote sensing. In the last two decades, a number of methods have been proposed to address the problem of hyperspectral data classification.In the present study, a structure based on learning capsule networks has been used to classify hyperspectral images, so that the network structure can have the most optimal generation of features by using a convolution layer and a capsule layer, and at the same time Avoid overfitting the on training data. The obtained results show the high quality of production features in the proposed structure.


With the development of remote sensing science, the use of hyperspectral images is becoming more widespread. Classification is one of the most popular topics in hyperspectral remote sensing. In the last two decades, a number of methods have been proposed to address the problem of hyperspectral data classification.In the present study, a structure based on learning capsule networks has been used to classify hyperspectral images, so that the network structure can have the most optimal generation of features by using a convolution layer and a capsule layer, and at the same time Avoid overfitting the on training data. The obtained results show the high quality of production features in the proposed structure.
In order to improve the classification accuracy, the feature extraction approach through the designed network and the classification by the Extreme Gradient Boosting was compared with the classification method by the global deep network. The proposed capsule approach consists of 3 basic layers: 1) Prime caps, which are capsules of size 8 and 32 with 9 × 9 filters and movement step 2, 2) Digitcaps with 10 16-dimensional capsules, and 3) fully connected layer. The results of examining two approaches for deep networking as well as combining capsule networks with XGBoost reinforcement tree algorithm were compared. Approaches such as SVM, RF-200, LSTM, GRU and GRU-Pretanh were considered to compare the proposed approach based on the configurations mentioned in their research.
Up in addition to the study and quality measurement of production vector deep features by the proposed method in different classifiers, the ability of deep global networks in the application of classification should also be examined. The results of examining two approaches for deep network and also combining CapsNet with XGBoost show that by using the proposed combined method, images are classified with 99% accuracy on training data and 97.5% accuracy on test data.

Up in addition to the study and quality measurement of production vector deep features by the proposed method in different classifiers, the ability of deep global networks in the application of classification should also be examined.

The results of examining two approaches for deep network and also combining CapsNet with XGBoost show that by using the proposed combined method, images are classified with 99% accuracy on training data and 97.5% accuracy on test data.

کلیدواژه‌ها [English]

  • Classification
  • Hyperspectral images
  • CapsNet
  • XGBoost
  • Fusion model
Akbari, D., Safari, A.R. & Homayouni, S., 2016, Improving Spectral-Spatial Classification of Supercritical Images by Using Spatial Information to Select Symbols, Scientific - Research Quarterly of Geographical Data (SEPEHR), 25(98), PP. 5-14.
Bengio, Y. & Ian, J., 2015, Goodfellow, and Aaron Courville. "Deep Learning", Nature, 521.7553, PP. 436-444.
Bruzzone, L., Chi, M. & Marconconi, M., 2006, A Novel Transductive SVM for Semisupervised Classification of Remote-Sensing Images, IEEE Transactions on Geoscience and Remote Sensing, 44(11), PP. 3363-3373.
Candemir, S., Jaeger, S., Palaniappan, K., Musco, J.P., Singh, R.K., Xue, Z. & McDonald, C.J., 2014, Lung Segmentation in Chest Radiographs Using Anatomical Atlases with Nonrigid Registration, IEEE Transactions on Medical Imaging, 33(2), PP. 577-590.
Camps-Valls, G., Gomez-Chova, L., Munoz-Mari, J., Vila-Frances, J. & Calpe-Maravilla, J., 2006, Composite Kernels for Hyperspectral Image Classification, IEEE Geoscience and Remote Sensing Letters, 3(1), PP. 93-97.
Camps-Valls, G., Shervashidze, N. & Borgwardt, K.M., 2010, Spatio-Spectral Remote Sensing Image Classification with Graph Kernels, IEEE Geoscience and Remote Sensing Letters, 7(4), PP. 741-745.
Chen, T. & Guestrin, C., 2016, Xgboost: A Scalable Tree Boosting System, In Proceedings of the 22nd ACM Sigkdd International Conference on Knowledge Discovery and Data Mining.
Dehghani, H., 2006, Classification of Distance Measurement Images with Large Dimensions and Limited Educational Examples, Ph.D. Thesis, Tarbiyat Modarres University, Department of Electronic Engineering, Tehran, Iran.
Domingos, P., 2012, A Few Useful Things to Know about Machine Learning, Communications of the ACM.
Du, P., Tan, K., Zhang, W. & Yan, Zh., 2008, ANN Classification of OMIS Hyperspectral Remotely Sensed Imagery: Experiments and Analysis, Congress on Image and Signal Processing, IEEE.
 
Fauvel, M., Chanussot, J., Benediktsson, J.A. & Sveinsson, J., 2007, Spectral and Spatial Classification of Hyperspectral Data Using SVMs and Morphological Profiles, IEEE Transactions on Geoscience and Remote Sensing, 46(11), PP. 2012-2020.
Fukunaga, A. & Sber, G., 2008, Providing an Optimal Method Based on Deep Learning for Spectral Classification Images with High Resolution Spatial Resolution in Semi-Urban Areas, Journal of Geomatics Science and Technology, 9(2), PP. 151-170.
Gao, Y., Gao, F., Dong, J. & Li, H.C., 2020, SAR Image Change Detection Based on Multiscale Capsule Network, IEEE Geosci. Remote Sens. Lett., 18(3).
Ghaffari, O., Voldan-Zoj, M.J. & Mokhtarzadeh, M., 2016, Selecting the Band in Order to Optimize the Spectral Separation of Supercritical Images, The 1st National Conference on Geospatial Information Technology, K.N. Toosi University of Technology, Tehran, Iran.
Ghamisi, P., Plaza, J., Chen, Y., Li, J. & Plaza, A.J., 2017, Advanced Spectral Classifiers for Hyperspectral Images: A Review, IEEE Geoscience and Remote Sensing Magazine, 5(1), PP. 8-32.
Ghassemian, H., Keshavarz, A. & Landgrebe, D., 2003, Hyper-Spectral Image Processing and Analyses, Space Magazine, 1(3), PP. 32-41.
Gualtieri, J.A. & Chetri, S.R., 2000, Support Vector Machines for Classification of Hyperspectral Data, Proc. IGARSS, Honolulu, _HI, PP. 813-815.
Gualtieri, J.A. & Cromp, R.F., 1999, Support Vector Machines for Hyperspectral Remote Sensing Classification, Proceedings Vol. 3584, 27th AIPR Workshop: Advances in Computer-Assisted Recognition.
Gualtieri, J.A., Chetri, S.R., Cromp, R.F. & Johnson, L.F., 1999, Support Vector Machines Classifiers as Applied to AVIRIS Data, In Summaries 8th JPL Airborne Earth Sience Workshop, JPL Pub., 99-17, PP. 217-227.
Hu, W., Huang, Y., Wei, L., Zhang, F. & Li, H., 2015, Deep Convolutional Neural Networks for Hyperspectral Image Classification, J. Sens, 2015, P. 258619.
Jimenez, L. & Landgrebe, D.A., 1999, Hyperspectral Data Analysis and Feature Reduction via Projection Pursuit, IEEE Trans. On Geoscience and Remote Sensing, 37(6), PP. 2653-2667.
Kaewpijit, S., Moigne, J.L., Ghazawi, T.E., 2003, Automatic Wavelet Spectral Analysis for Reduction of Hyperspectral Imagery, IEEE Trans. on Geoscience and Remote Sensing, 41(4), PP. 863-871.
Keshavarz, A. & Ghasemiyan, H., 2005, A High-Speed Vector Machine-Based Algorithm for Classifying Hayperspectral Images Using Spatial Correlation, Iranian Journal of Electrical and Computer Engineering, 3(1), PP. 37-44.
Landgrebe, D.A., 2002, Hyperspectral Image Data Analysis, IEEE Signal Processing Magazine, 19(1), PP. 17-28.
LeCun, Y., Bengio, Y. & Hinton, G., 2015, Deep Learning, Nature, 521, PP. 436-444.
Lee, C. & Landgrebe, D.A., 1993, Feature Extraction Based on Decision Boundaries, IEEE Trans. on Pattern Analysis and Machine Intelligence, 15(4), PP. 388-400.
Li, H., 2018, Deep Learning for Natural Language Processing: Advantages and Challenges, National Science Review., 5(1), PP. 24-26.
Liyang, W., Yongyi, Y., Nishikawa, R.M., Wernick, M.N. & Edwards, A., 2005, Relevance Vector Machine for Automatic Detection of Clustered Micro Calcifications, IEEE Trans. Med. Imag., 24(10), PP. 1278-1285.
Luo, Y., Zou, J., Yao, C., Li, T. & Bai, G., 2018, HSI-CNN: A Novel Convolution Neural Network for Hyperspectral Image., arXiv 2018, arXiv:1802.10478.
Maggiori, E., Tarabalka, Y., Charpiat, G. & Alliez, P., 2016, High-Resolution Semantic Labeling with Convolutional Neural Networks, ArXiv: 1611.01962.
Melgani, F. & Bruzzone, L., 2004, Classification of Hyperspectral Remote Sensing Images with Support Vector Machines, IEEE Transactions on Geoscience and Remote Sensing, 42(8), PP.1778-1790.
Paoletti, M.E., Plaza, J. & Plaza, A., 2018, A New Deep Convolutional Neural Network for Fast Hyperspectral Image Classification, ISPRS Journal of Photogrammetry and Remote Sensing, 145, PP. 120-147. doi:10.1016/ j.isprsjprs.2017.11.021.
Paoletti, M.E., Haut, J.M., Fernandez-Beltran, R., Plaza, J., Plaza, A., Jun, L. & Pla, F., 2019, Capsul Networks for Hyperspectral Image Classification, IEEE Transactions on Geoscience and Remote Sensing, 57(4), PP. 2145-2160. doi:10.1109/TGRS.2018.2871782.
Sabour, S., Frosst, N. & Hinton, G.E., 2017, Dynamic Routing between Capsules, In Proceedings of the Advances in Neural Information Processing Systems, Long Beach, CA, USA, 4-9 December 2017, PP. 3859-3869.
Shahhoseeini, R., 2009, Evaluation of Support Vector Machines in the Classification of Hyperspectral Remote Sensing Data, Ms.c Thesis, University of Tehran, Tehran, Iran.
Wang, Y., Sun, A., Han, J., Liu, Y. & Zhu, X., 2018, Sentiment Analysis by Capsules, In Proceedings of the 2018 World Wide Web Conference on World Wide Web, Lyon, France, 23-27 April 2018; PP. 1165-1174.
Xu, J.L., Esquerre, C. & Sun, D.W., 2018. Methods for Performing Dimensionality Reduction in Hyperspectral Image Classification, Journal of Near Infrared Spectroscopy, 26(1), PP. 61-75.
Xu, Q., Wang, D.Y. & Luo, B., 2020, Faster Multiscale Capsule Network with Octave Convolution for Hyperspectral Image Classification, IEEE Geosci. Remote Sens. Lett, 18(2).
Xue, Z., You, D., Candemir, S., Jaeger, S., Antani, S., Long, L.R. & Thoma, G.R., 2015, Chest X-Ray Image View Classification, In Proceedings of the 28th International Symposium on Computer-Based Medical Systems (CBMS) (PP. 66-71), Brazil: Ribeião Preto.
Yu, Y., Gu, T., Guan, H., Li, D. & Jin, S., 2019, Vehicle Detection from High-Resolution Remote Sensing Imagery Using Convolutional Capsule Networks, IEEE Geosci. Remote Sens. Lett., 16, PP. 1894-1898.