Flood contributes a key role in devastating natural and man-made areas. Floods usually are occurred when there is a considerable number of clouds in the sky making optic data useless. Synthetic aperture radar (SAR) images can be a valuable data source in earth observation tasks. The most important characteristic of the radar image is its ability to penetrate the cloud and dust. Therefore, monitoring earth in cloudy or rainy weather can be available by this kind of dataset. In the last few years by improving machine learning methods and development of convolutional neural networks in remote sensing applications we are facing with extremely high improvement in classification tasks. In this paper, we use dual-polarized VV and VH backscatter values of Sentinel-1 and Shuttle Radar Topography Mission (SRTM) digital elevation model (DEM) dataset in a proposed convolutional neural network to generate a land cover map of a flooded area before and after happening. Obtained classification results vary between 93.3% to 98.5% for different training sizes. By comparing the generated classified maps, flooded areas of each class can be extracted.
ABSTRACT:The goal of ensemble learning methods like Bagging and Boosting is to improve the classification results of some weak classifiers gradually. Usually, Boosting algorithms show better results than Bagging. In this article, we have examined the possibility of fusion of non-thermal and thermal bands of Landsat 8 satellite images for cloud detection by using the boosting method. We used SVM as a base learner and the performance of two kinds of Boosting methods including AdaBoost.M1 and σ Boost was compared on remote sensing images of Landsat 8 satellite. We first extracted the co-occurrence matrix features of non-thermal and thermal bands separately and then used PCA method for feature selection. In the next step AdaBoost.M1 and σ Boost algorithms were applied on non-thermal and thermal bands and finally, the classifiers were fused using majority voting. Also, we showed that by changing the regularization parameter (C) the result of σ Boost algorithm can significantly change and achieve overall accuracy and cloud producer accuracy of 74%, and 0.53 kappa coefficient that shows better results in comparison to AdaBoost.M1.
Accurate building mapping using remote sensing data is challenging because of the complexity of building structures, particularly in populated cities. LiDAR data are widely used for building extraction because they provide height information, which can help distinguish buildings from other tall objects. However, tall trees and bridges in the vicinity of buildings can limit the application of LiDAR data, particularly in urban areas. Combining LiDAR and orthoimages can help in such situations, because orthoimages can provide information on the physical properties of objects, such as reflectance characteristics. One efficient way to combine these two data sources is to use convolutional neural networks (CNN). This study proposes a CNN architecture based on dense attention blocks for building detection in southern Toronto and Massachusetts. The stacking of information from multiple previous layers was inspired by dense attention networks (DANs). DAN blocks consist of batch normalization, convolution, dropout, and average pooling layers to extract high- and low-level features. Compared with two other widely used deep learning techniques, VGG16 and Resnet50, the proposed method has a simpler architecture and converges faster with higher accuracy. In addition, a comparison with the two other state-of-the-art deep learning methods, including U-net and ResUnet, showed that our proposed technique could achieve a higher F1-score, of 0.71, compared with 0.42 for U-net and 0.49 for ResUnet.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.