Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor Data
With the development of high-resolution optical sensors, the classification of ground objects combined with multivariate optical sensors is a hot topic at present. Deep learning methods, such as convolutional neural networks, are applied to feature extraction and classification. In this work, a nove...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2019-01-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | http://www.mdpi.com/1424-8220/19/1/204 |
| _version_ | 1828150326717317120 |
|---|---|
| author | Chenming Li Yongchang Wang Xiaoke Zhang Hongmin Gao Yao Yang Jiawei Wang |
| author_facet | Chenming Li Yongchang Wang Xiaoke Zhang Hongmin Gao Yao Yang Jiawei Wang |
| author_sort | Chenming Li |
| collection | DOAJ |
| description | With the development of high-resolution optical sensors, the classification of ground objects combined with multivariate optical sensors is a hot topic at present. Deep learning methods, such as convolutional neural networks, are applied to feature extraction and classification. In this work, a novel deep belief network (DBN) hyperspectral image classification method based on multivariate optical sensors and stacked by restricted Boltzmann machines is proposed. We introduced the DBN framework to classify spatial hyperspectral sensor data on the basis of DBN. Then, the improved method (combination of spectral and spatial information) was verified. After unsupervised pretraining and supervised fine-tuning, the DBN model could successfully learn features. Additionally, we added a logistic regression layer that could classify the hyperspectral images. Moreover, the proposed training method, which fuses spectral and spatial information, was tested over the Indian Pines and Pavia University datasets. The advantages of this method over traditional methods are as follows: (1) the network has deep structure and the ability of feature extraction is stronger than traditional classifiers; (2) experimental results indicate that our method outperforms traditional classification and other deep learning approaches. |
| first_indexed | 2024-04-11T21:42:52Z |
| format | Article |
| id | doaj.art-3e83762776614a678fa06b0bf8ae1ac9 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-11T21:42:52Z |
| publishDate | 2019-01-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-3e83762776614a678fa06b0bf8ae1ac92022-12-22T04:01:32ZengMDPI AGSensors1424-82202019-01-0119120410.3390/s19010204s19010204Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor DataChenming Li0Yongchang Wang1Xiaoke Zhang2Hongmin Gao3Yao Yang4Jiawei Wang5College of Computer and Information, Hohai University, Nanjing 211100, ChinaCollege of Computer and Information, Hohai University, Nanjing 211100, ChinaSchool of Public Administration, Hohai University, Nanjing 211100, ChinaCollege of Computer and Information, Hohai University, Nanjing 211100, ChinaCollege of Computer and Information, Hohai University, Nanjing 211100, ChinaCollege of Computer and Information, Hohai University, Nanjing 211100, ChinaWith the development of high-resolution optical sensors, the classification of ground objects combined with multivariate optical sensors is a hot topic at present. Deep learning methods, such as convolutional neural networks, are applied to feature extraction and classification. In this work, a novel deep belief network (DBN) hyperspectral image classification method based on multivariate optical sensors and stacked by restricted Boltzmann machines is proposed. We introduced the DBN framework to classify spatial hyperspectral sensor data on the basis of DBN. Then, the improved method (combination of spectral and spatial information) was verified. After unsupervised pretraining and supervised fine-tuning, the DBN model could successfully learn features. Additionally, we added a logistic regression layer that could classify the hyperspectral images. Moreover, the proposed training method, which fuses spectral and spatial information, was tested over the Indian Pines and Pavia University datasets. The advantages of this method over traditional methods are as follows: (1) the network has deep structure and the ability of feature extraction is stronger than traditional classifiers; (2) experimental results indicate that our method outperforms traditional classification and other deep learning approaches.http://www.mdpi.com/1424-8220/19/1/204hyperspectral imagedeep learningfeature extractionclassificationremote sensorsmulti-sensor fusion |
| spellingShingle | Chenming Li Yongchang Wang Xiaoke Zhang Hongmin Gao Yao Yang Jiawei Wang Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor Data Sensors hyperspectral image deep learning feature extraction classification remote sensors multi-sensor fusion |
| title | Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor Data |
| title_full | Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor Data |
| title_fullStr | Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor Data |
| title_full_unstemmed | Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor Data |
| title_short | Deep Belief Network for Spectral–Spatial Classification of Hyperspectral Remote Sensor Data |
| title_sort | deep belief network for spectral spatial classification of hyperspectral remote sensor data |
| topic | hyperspectral image deep learning feature extraction classification remote sensors multi-sensor fusion |
| url | http://www.mdpi.com/1424-8220/19/1/204 |
| work_keys_str_mv | AT chenmingli deepbeliefnetworkforspectralspatialclassificationofhyperspectralremotesensordata AT yongchangwang deepbeliefnetworkforspectralspatialclassificationofhyperspectralremotesensordata AT xiaokezhang deepbeliefnetworkforspectralspatialclassificationofhyperspectralremotesensordata AT hongmingao deepbeliefnetworkforspectralspatialclassificationofhyperspectralremotesensordata AT yaoyang deepbeliefnetworkforspectralspatialclassificationofhyperspectralremotesensordata AT jiaweiwang deepbeliefnetworkforspectralspatialclassificationofhyperspectralremotesensordata |