scholarly journals LAND COVER CLASSIFICATION USING HIGH RESOLUTION SATELLITE IMAGE BASED ON DEEP LEARNING

2020 ◽  
Vol XLII-3/W10 ◽  
pp. 685-690
Author(s):  
M. Zhu ◽  
B. Wu ◽  
Y. N. He ◽  
Y. Q. He
Keyword(s):  
Remote Sensing ◽  
Deep Learning ◽  
High Resolution ◽  
Land Cover ◽  
Satellite Image ◽  
Remote Sensing Data ◽  
Semantic Segmentation ◽  
Land Cover Classification ◽  
Main Task ◽  
High Resolution Satellite Image

Abstract. In the coming era of big data, the high resolution satellite image plays an important role in providing a rich source of information for a variety of applications. Land cover classification is a major field of remote sensing application. The main task of land cover classification is to divide the pixels or regions in remote sensing imagery into several categories according to application requirements. Recently, machine interpretation methods including artificial neural network and decision tree are developing rapidly with certain fruits achieved. Compared with traditional methods, deep learning is completely data-driven, which can automatically find the best ways to extract land cover features through high resolution satellite image. This study presents a detailed investigation of convolutional neural networks for the classification of complex land cover classes using high resolution satellite image. The main contributions of this paper are as follows: (1) Aiming at the uneven spatial distribution of surface coverage, we study the training errors caused by this uneven distribution. An improved SMOTE algorithm is designed for automatic processing the task of sample augmentation. Through experimental verification, the improver algorithm can increase 2–5% classification accuracy by the same network structure. (2) The main representations of the network are also shared between the edge loss reinforced structures and semantic segmentation, which means that the CNN simultaneously achieves semantic segmentation by edge detection. (3) We use Beijing-2 satellite (BJ-2) remote sensing data for training and evaluation with Integrated Model, and the total accuracy reaches 89.6%.

scholarly journals How Well Do Deep Learning-Based Methods for Land Cover Classification and Object Detection Perform on High Resolution Remote Sensing Imagery?

2020 ◽  
Vol 12 (3) ◽  
pp. 417 ◽  
Author(s):  
Xin Zhang ◽  
Liangxiu Han ◽  
Lianghao Han ◽  
Liang Zhu
Keyword(s):  
Remote Sensing ◽  
Deep Learning ◽  
High Resolution ◽  
Land Cover ◽  
Object Detection ◽  
Spatial Resolution ◽  
High Spatial Resolution ◽  
Remote Sensing Data ◽  
Land Cover Classification ◽  
Sensing Data

Land cover information plays an important role in mapping ecological and environmental changes in Earth’s diverse landscapes for ecosystem monitoring. Remote sensing data have been widely used for the study of land cover, enabling efficient mapping of changes of the Earth surface from Space. Although the availability of high-resolution remote sensing imagery increases significantly every year, traditional land cover analysis approaches based on pixel and object levels are not optimal. Recent advancement in deep learning has achieved remarkable success on image recognition field and has shown potential in high spatial resolution remote sensing applications, including classification and object detection. In this paper, a comprehensive review on land cover classification and object detection approaches using high resolution imagery is provided. Through two case studies, we demonstrated the applications of the state-of-the-art deep learning models to high spatial resolution remote sensing data for land cover classification and object detection and evaluated their performances against traditional approaches. For a land cover classification task, the deep-learning-based methods provide an end-to-end solution by using both spatial and spectral information. They have shown better performance than the traditional pixel-based method, especially for the categories of different vegetation. For an objective detection task, the deep-learning-based object detection method achieved more than 98% accuracy in a large area; its high accuracy and efficiency could relieve the burden of the traditional, labour-intensive method. However, considering the diversity of remote sensing data, more training datasets are required in order to improve the generalisation and the robustness of deep learning-based models.

scholarly journals Dual Attention Feature Fusion and Adaptive Context for Accurate Segmentation of Very High-Resolution Remote Sensing Images

2021 ◽  
Vol 13 (18) ◽  
pp. 3715
Author(s):  
Hao Shi ◽  
Jiahe Fan ◽  
Yupei Wang ◽  
Liang Chen
Keyword(s):  
Remote Sensing ◽  
High Resolution ◽  
Land Cover ◽  
Feature Fusion ◽  
Semantic Segmentation ◽  
Land Cover Classification ◽  
Contextual Cues ◽  
Remote Sensing Images ◽  
Object Boundary ◽  
Convolutional Network

Land cover classification of high-resolution remote sensing images aims to obtain pixel-level land cover understanding, which is often modeled as semantic segmentation of remote sensing images. In recent years, convolutional network (CNN)-based land cover classification methods have achieved great advancement. However, previous methods fail to generate fine segmentation results, especially for the object boundary pixels. In order to obtain boundary-preserving predictions, we first propose to incorporate spatially adapting contextual cues. In this way, objects with similar appearance can be effectively distinguished with the extracted global contextual cues, which are very helpful to identify pixels near object boundaries. On this basis, low-level spatial details and high-level semantic cues are effectively fused with the help of our proposed dual attention mechanism. Concretely, when fusing multi-level features, we utilize the dual attention feature fusion module based on both spatial and channel attention mechanisms to relieve the influence of the large gap, and further improve the segmentation accuracy of pixels near object boundaries. Extensive experiments were carried out on the ISPRS 2D Semantic Labeling Vaihingen data and GaoFen-2 data to demonstrate the effectiveness of our proposed method. Our method achieves better performance compared with other state-of-the-art methods.

scholarly journals Estimation of Boreal Forest Growing Stock Volume in Russia from Sentinel-2 MSI and Land Cover Classification

2021 ◽  
Vol 13 (21) ◽  
pp. 4483
Author(s):  
W. Gareth Rees ◽  
Jack Tomaney ◽  
Olga Tutubalina ◽  
Vasily Zharko ◽  
Sergey Bartalev
Keyword(s):  
Remote Sensing ◽  
Land Cover ◽  
Boreal Forest ◽  
Remote Sensing Data ◽  
Land Cover Classification ◽  
Fundamental Parameter ◽  
Sensing Data ◽  
Growing Stock ◽  
Stock Volume ◽  
Sentinel 2

Growing stock volume (GSV) is a fundamental parameter of forests, closely related to the above-ground biomass and hence to carbon storage. Estimation of GSV at regional to global scales depends on the use of satellite remote sensing data, although accuracies are generally lower over the sparse boreal forest. This is especially true of boreal forest in Russia, for which knowledge of GSV is currently poor despite its global importance. Here we develop a new empirical method in which the primary remote sensing data source is a single summer Sentinel-2 MSI image, augmented by land-cover classification based on the same MSI image trained using MODIS-derived data. In our work the method is calibrated and validated using an extensive set of field measurements from two contrasting regions of the Russian arctic. Results show that GSV can be estimated with an RMS uncertainty of approximately 35–55%, comparable to other spaceborne estimates of low-GSV forest areas, with 70% spatial correspondence between our GSV maps and existing products derived from MODIS data. Our empirical approach requires somewhat laborious data collection when used for upscaling from field data, but could also be used to downscale global data.

scholarly journals Leveraging machine learning to produce a cost-effective national building height map of Ireland

2021 ◽  
Author(s):  
Eoghan Keany ◽  
Geoffrey Bessardon ◽  
Emily Gleeson
Keyword(s):  
Machine Learning ◽  
Remote Sensing ◽  
High Resolution ◽  
Land Cover ◽  
Urban Areas ◽  
Cost Effective ◽  
Land Cover Classification ◽  
National Scale ◽  
Building Height ◽  
Height Data

<p>To represent surface thermal, turbulent and humidity exchanges, Numerical Weather Prediction (NWP) systems require a land-cover classification map to calculate sur-face parameters used in surface flux estimation. The latest land-cover classification map used in the HARMONIE-AROME configuration of the shared ALADIN-HIRLAMNWP system for operational weather forecasting is ECOCLIMAP-SG (ECO-SG). The first evaluation of ECO-SG over Ireland suggested that sparse urban areas are underestimated and instead appear as vegetation areas (1). While the work of (2) on land-cover classification helps to correct the horizontal extent of urban areas, the method does not provide information on the vertical characteristics of urban areas. ECO-SG urban classification implicitly includes building heights (3), and any improvement to ECO-SG urban area extent requires a complementary building height dataset.</p><p>Openly accessible building height data at a national scale does not exist for the island of Ireland. This work seeks to address this gap in availability by extrapolating the preexisting localised building height data across the entire island. The study utilises information from both the temporal and spatial dimensions by creating band-wise temporal aggregation statistics from morphological operations, for both the Sentinel-1A/B and Sentinel-2A/B constellations (4). The extrapolation uses building height information from the Copernicus Urban Atlas, which contains regional coverage for Dublin at 10 m x10 m resolution (5). Various regression models were then trained on these aggregated statistics to make pixel-wise building height estimates. These model estimates were then evaluated with an adjusted RMSE metric, with the most accurate model chosen to map the entire country. This method relies solely on freely available satellite imagery and open-source software, providing a cost-effective mapping service at a national scale that can be updated more frequently, unlike expensive once-off private mapping services. Furthermore, this process could be applied by these services to reduce costs by taking a small representative sample and extrapolating the rest of the area. This method can be applied beyond national borders providing a uniform map that does not depends on the different private service practices facilitating the updates of global or continental land-cover information used in NWP.</p><p> </p><p>(1) G. Bessardon and E. Gleeson, “Using the best available physiography to improve weather forecasts for Ireland,” in Challenges in High-Resolution Short Range NWP at European level including forecaster-developer cooperation, European Meteorological Society, 2019.</p><p>(2) E. Walsh, et al., “Using machine learning to produce a very high-resolution land-cover map for Ireland, ” Advances in Science and Research,  (accepted for publication).</p><p>(3) CNRM, "Wiki - ECOCLIMAP-SG" https://opensource.umr-cnrm.fr/projects/ecoclimap-sg/wiki</p><p>(4) D. Frantz, et al., “National-scale mapping of building height using sentinel-1 and sentinel-2 time series,” Remote Sensing of Environment, vol. 252, 2021.</p><p>(5) M. Fitrzyk, et al., “Esa Copernicus sentinel-1 exploitation activities,” in IGARSS 2019-2019 IEEE International Geoscience and Remote Sensing Symposium, IEEE, 2019.</p>

Sign in / Sign up

Export Citation Format

Share Document