Fully Connected Conditional Radom Fields for High Resolution Remote Sensing Land Use/Land Cover Classification with Convolutional Neural Networks

2018 ◽  
Author(s):  
Bin Zhang ◽  
Cunpeng Wang ◽  
Yonglin Shen ◽  
Yueyan Liu
Keyword(s):  
Remote Sensing ◽  
Neural Networks ◽  
Land Use ◽  
High Resolution ◽  
Svm Classifier ◽  
Land Use Land Cover ◽  
Remote Sensing Images ◽  
Feature Descriptors ◽  
New Feature ◽  
Fully Connected

The interpretation of land use/land cover (LULC) is a hotspot and difficult issue in the field of high resolution remote sensing image processing as well as land resource management. Training a new (or existing) Convolutional Neural Networks (CNNs) architecture fully for LULC classification needs a large amount of remote sensing images. Thus, fine-tuning a pre-trained CNNs for LULC is acceptable. To improve the classification accuracy for high resolution remote sensing images, it is necessary to utilize some hand-crafted features and adopt a classifier for post-processing. A Fully Connected Conditional Radom Fields (FC-CRFs), to utilize the fine-tuned CNNs layers, hand-crafted features and fully connected pairwise potentials, is proposed for image classification of high resolution remote sensing images. First, an existing CNNs model is adopted, and the parameters of CNNs are fine-tuned by training datasets. Then, the probabilities of image pixels belong to each class type also could be calculated. Second, we consider the hand-craft features, combined with Support Vector Machine (SVM) classifier, the probabilities belong to each LULC class type are achieved. Combined with the probabilities achieved by fine-tuned CNNs, new feature descriptors are built. Finally, FC-CRFs are introduced to get the classification results, while the unary potentials are achieved by the new feature descriptors and SVM classifier, and the pairwise potentials are achieved by the hand-crafted features. Experimental results show that the proposed classification scheme can achieve impressive performance when the total accuracy reached to about 85%.

scholarly journals Fully Connected Conditional Random Fields for High-Resolution Remote Sensing Land Use/Land Cover Classification with Convolutional Neural Networks

2018 ◽  
Vol 10 (12) ◽  
pp. 1889 ◽  
Author(s):  
Bin Zhang ◽  
Cunpeng Wang ◽  
Yonglin Shen ◽  
Yueyan Liu
Keyword(s):  
Remote Sensing ◽  
High Resolution ◽  
Random Fields ◽  
Conditional Random Fields ◽  
Svm Classifier ◽  
Spectral Features ◽  
Remote Sensing Images ◽  
Feature Descriptors ◽  
New Feature ◽  
Fully Connected

The interpretation of land use and land cover (LULC) is an important issue in the fields of high-resolution remote sensing (RS) image processing and land resource management. Fully training a new or existing convolutional neural network (CNN) architecture for LULC classification requires a large amount of remote sensing images. Thus, fine-tuning a pre-trained CNN for LULC detection is required. To improve the classification accuracy for high resolution remote sensing images, it is necessary to use another feature descriptor and to adopt a classifier for post-processing. A fully connected conditional random fields (FC-CRF), to use the fine-tuned CNN layers, spectral features, and fully connected pairwise potentials, is proposed for image classification of high-resolution remote sensing images. First, an existing CNN model is adopted, and the parameters of CNN are fine-tuned by training datasets. Then, the probabilities of image pixels belong to each class type are calculated. Second, we consider the spectral features and digital surface model (DSM) and combined with a support vector machine (SVM) classifier, the probabilities belong to each LULC class type are determined. Combined with the probabilities achieved by the fine-tuned CNN, new feature descriptors are built. Finally, FC-CRF are introduced to produce the classification results, whereas the unary potentials are achieved by the new feature descriptors and SVM classifier, and the pairwise potentials are achieved by the three-band RS imagery and DSM. Experimental results show that the proposed classification scheme achieves good performance when the total accuracy is about 85%.

scholarly journals Fully Connected Conditional Random Fields for High-Resolution Remote Sensing Land Use/Land Cover Classification with Convolutional Neural Networks

2018 ◽  
Author(s):  
Bin Zhang ◽  
Cunpeng Wang ◽  
Yonglin Shen ◽  
Yueyan Liu
Keyword(s):  
Remote Sensing ◽  
High Resolution ◽  
Random Fields ◽  
Conditional Random Fields ◽  
Svm Classifier ◽  
Spectral Features ◽  
Remote Sensing Images ◽  
Feature Descriptors ◽  
New Feature ◽  
Fully Connected

The interpretation of land use and land cover (LULC) is an important issue in the fields of high-resolution remote sensing (RS) image processing and land resource management. Fully training a new or existing convolutional neural network (CNN) architecture for LULC classification requires a large amount of remote sensing images. Thus, fine-tuning a pre-trained CNN for LULC detection is required. To improve the classification accuracy for high resolution remote sensing images, it is necessary to use another feature descriptor and to adopt a classifier for post-processing. A fully connected conditional random fields (FC-CRF), to use the fine-tuned CNN layers, spectral features, and fully connected pairwise potentials, is proposed for image classification of high-resolution remote sensing images. First, an existing CNN model is adopted, and the parameters of CNN are fine-tuned by training datasets. Then, the probabilities of image pixels belong to each class type are calculated. Second, we consider the spectral features and digital surface model (DSM) and combined with a support vector machine (SVM) classifier, the probabilities belong to each LULC class type are determined. Combined with the probabilities achieved by the fine-tuned CNN, new feature descriptors are built. Finally, FC-CRF are introduced to produce the classification results, whereas the unary potentials are achieved by the new feature descriptors and SVM classifier, and the pairwise potentials are achieved by the three-band RS imagery and DSM. Experimental results show that the proposed classification scheme achieves good performance when the total accuracy is about 85%.

Discussion on Remote Dynamic Sensing Monitoring Method for Land in Regions under Rapid Economic Development

2012 ◽  
Vol 518-523 ◽  
pp. 5788-5792
Author(s):  
Zheng Dong Xie ◽  
Jian Zhang ◽  
Bu Zhuo Peng
Keyword(s):  
Remote Sensing ◽  
Land Use ◽  
High Resolution ◽  
Property Right ◽  
Remote Sensing Images ◽  
Monitoring Method ◽  
Nanjing City ◽  
Platform System ◽  
First Time

The paper was supported by The Second Land Investigation Item and took Nanjing city, Jiangsu Province as a case study. The research of the theory, technique and application for land use investigation was achieved by the high-resolution remote sensing images for application, designed a set of technique of land use investigation for land property right management. The database and platform system were established to carry out the dynamic management of land use. Based on the summarization of the correlative studies, The paper designed a set of technique of land investigation for land property right management and also designed the technical process, dealt with the remote sensing images, detected the changed information, classified the land, investigated the land property right and established the database to serve for the management of land property right. And it has been successfully used in Nanjing. It’s unique to use the high-resolution remote sensing images by QuichBird for the scale of 1:5000 in land use investigation in area cities which is also the first time in Nanjing City.

scholarly journals A Neural Network-Based Distributed Compressed Sensing Algorithm to Derive Continuous Hyperspectral Images in Wireless Sensor Networks

2022 ◽  
Author(s):  
Md. Sarkar Hasanuzzaman
Keyword(s):  
Neural Network ◽  
Remote Sensing ◽  
Neural Networks ◽  
Wireless Sensor Networks ◽  
Sensor Networks ◽  
High Resolution ◽  
Super Resolution ◽  
Hyperspectral Images ◽  
Wireless Sensor ◽  
Remote Sensing Images

Abstract Hyperspectral imaging is a versatile and powerful technology for gathering geo-data. Planes and satellites equipped with hyperspectral cameras are currently the leading contenders for large-scale imaging projects. Aiming at the shortcomings of traditional methods for detecting sparse representation of multi-spectral images, this paper proposes wireless sensor networks (WSNs) based single-hyperspectral image super-resolution method based on deep residual convolutional neural networks. We propose a different strategy that involves merging cheaper multispectral sensors to achieve hyperspectral-like spectral resolution while maintaining the WSN's spatial resolution. This method studies and mines the nonlinear relationship between low-resolution remote sensing images and high-resolution remote sensing images, constructs a deep residual convolutional neural network, connects multiple residual blocks in series, and removes some unnecessary modules. For this purpose, a decision support system is used that provides the outcome to the next layer. Finally, this paper, fully explores the similarities between natural images and hyperspectral images, use natural image samples to train convolutional neural networks, and further use migration learning to introduce the trained network model to the super-resolution problem of high-resolution remote sensing images, and solve the lack of training samples problem. A comparison between different algorithms for processing data on datasets collected in situ and via remote sensing is used to evaluate the proposed approach. The experimental results show that the method has good performance and can obtain better super-resolution effects.

Sign in / Sign up

Export Citation Format

Share Document