High-resolution Remote Sensing Image Data Research Articles

PMNet: a multi-branch and multi-scale semantic segmentation approach to water extraction from high-resolution remote sensing images with edge-cloud computing

In the field of remote sensing image interpretation, automatically extracting water body information from high-resolution images is a key task. However, facing the complex multi-scale features in high-resolution remote sensing images, traditional methods and basic deep convolutional neural networks are difficult to effectively capture the global spatial relationship of the target objects, resulting in incomplete, rough shape and blurred edges of the extracted water body information. Meanwhile, massive image data processing usually leads to computational resource overload and inefficiency. Fortunately, the local data processing capability of edge computing combined with the powerful computational resources of cloud centres can provide timely and efficient computation and storage for high-resolution remote sensing image segmentation. In this regard, this paper proposes PMNet, a lightweight deep learning network for edge-cloud collaboration, which utilises a pipelined multi-step aggregation method to capture image information at different scales and understand the relationships between remote pixels through horizontal and vertical spatial dimensions. Also, it adopts a combination of multiple decoding branches in the decoding stage instead of the traditional single decoding branch. The accuracy of the results is improved while reducing the consumption of system resources. The model obtained F1-score of 90.22 and 88.57 on Landsat-8 and GID remote sensing image datasets with low model complexity, which is better than other semantic segmentation models, highlighting the potential of mobile edge computing in processing massive high-resolution remote sensing image data.

Quantitative effect of lithologic-geomorphic coupling on gully distribution at the regional scale: An example of the Dry Valleys (SW China)

Gully erosion is an important soil erosion type that destroys the surface structure and causes severe soil and water loss. Currently, the quantitative relationship between the spatial distribution of gullies and lithologic-geomorphic coupling is unclear. Gullies were interpreted by using high-resolution remote sensing image data to establish a database. Correlation analysis and geographic information technology were used to quantify the correlation between the gully distribution and lithologic-geomorphic types. The results show that the gully spatial distribution is characterized by clustering, and the lithologic-geomorphic type has a large effect on the spatial distribution of gullies. The coupling zone of phyllites and schists and intermediate-relief and intermediate-altitude mountains (P-II) has the greatest clustering; the zone of loose Quaternary sediments and intermediate-altitude plains (L-IP) has the least clustering. The lithologic-geomorphic coupling affects the initiation, development, and distribution of gullies; the more fragile the lithology and the greater the relief amplitude, the easier it is for gullies to form. This study will make it possible to better understand the distribution pattern and formation mechanism of gullies at a regional scale, and it will also provide a scientific reference for regional gully management.

Landslide Extraction from High-Resolution Remote Sensing Imagery Using Fully Convolutional Spectral–Topographic Fusion Network

Considering the complexity of landslide hazards, their manual investigation lacks efficiency and is time-consuming, especially in high-altitude plateau areas. Therefore, extracting landslide information using remote sensing technology has great advantages. In this study, comprehensive research was carried out on the landslide features of high-resolution remote sensing images on the Mangkam dataset. Based on the idea of feature-driven classification, the landslide extraction model of a fully convolutional spectral–topographic fusion network (FSTF-Net) based on a deep convolutional neural network of multi-source data fusion is proposed, which takes into account the topographic factor (slope and aspect) and the normalized difference vegetation index (NDVI) as multi-source data input by which to train the model. In this paper, a high-resolution remote sensing image classification method based on a fully convolutional network was used to extract the landslide information, thereby realizing the accurate extraction of the landslide and surrounding ground-object information. With Mangkam County in the southeast of the Qinghai–Tibet Plateau China as the study area, the proposed method was evaluated based on the high-precision digital elevation model (DEM) generated from stereoscopic images of Resources Satellite-3 and multi-source high-resolution remote sensing image data (Beijing-2, Worldview-3, and SuperView-1). Results show that our method had a landslide detection precision of 0.85 and an overall classification accuracy of 0.89. Compared with the latest DeepLab_v3+, our model increases the landslide detection precision by 5%. Thus, the proposed FSTF-Net model has high reliability and robustness.

Exploring the Rapid Assessment Method for Nature Reserve Landscape Protection Effectiveness—A Case Study of Liancheng National Nature Reserve, Gansu, China

The rapid assessment of the effectiveness of landscape protection in nature reserves is of great significance for the scientific formulation of protection and management countermeasures. There is also an urgent problem to be solved for the construction and management of nature reserves in China. Using high-resolution remote sensing image data, this study analyzes the landscape dynamics in the Liancheng National Nature Reserve (LNNR) and their driving factors since the reserve’s promotion to the national level in 2005, and proposes a comprehensive evaluation method for the effectiveness of landscape protection in protected areas based on the Landscape Transfer Index (LTI), Protected Landscape Integrity Index (PLII), and Interfered Landscape Sprawl Index (ILSI). Between 2006 and 2019, the area of protected landscape—namely woodland, grassland, and water—in the LNNR decreased, while the area of interfered landscape such as residential land, industrial and mining land, and water conservancy facility land increased. The LTI was −0.14, and among the driving factors, the development of industry and mining, land use by indigenous inhabitants, and the development of the transport industry made the highest contributions to the landscape transfer tendency, respectively 34.79%, 28.98%, and 17.30%. The results indicate that the overall quality of the landscape declined slightly during this period, mainly as a result of industrial and mining development, indigenous use of the land, and road construction. The PLII decreased from 82.7 to 68.7; this result indicates that the spatial pattern of the protected landscape became more fragmented, and the degree of human interference in the landscape increased. The ILSI increased from 26.61 to 26.68 which indicates that the scope of influence of human interference did not change significantly. The effectiveness of landscape protection in the LNNR is low. Despite the insignificant nature of these changes, they still require attention and timely remedial measures. The methodology proposed in this study may be applicable to the rapid assessment of the effectiveness of landscape conservation in various types of nature conservation sites around the world.

Deciphering the Link Between Mental Health and Green Space in Shenzhen, China: The Mediating Impact of Residents' Satisfaction.

Rapid urbanization and increasing urban density in China threaten residents' mental health. As a vital component of built environments, green space plays a key role in individuals' psychological well-being; however, the mediating effect of residents' satisfaction with the green space environment on the relationship between urban greening and residents' mental health in Chinese contexts has yet to be thoroughly explored. To fill this knowledge gap, this paper attempts to reveal the internal logic and mechanism underlying the linkages between green space, residents' mental health, and their satisfaction with green space in Shenzhen, China. Specifically, this paper explores the mediating role of residents' satisfaction with a green space environment using questionnaire survey data, “Quick Bird-2” high-resolution remote sensing image data, and a multilevel regression model. Our empirical findings indicate that the relative range of neighboring green spaces can directly improve residents' mental health. More importantly, the relationship between the relative scope of green space and residents' mental health is mediated by residents' satisfaction with the green space environment rather than its direct health effects. Given the influence of green space on residents' satisfaction with the environment, green space indirectly affects mental health. These findings should provide the government useful guidance for considering the spatial distribution and quantity of green space. Our results should also help residents improve their actual experiences and subjective satisfaction with the green space environment.

AFNet: Adaptive Fusion Network for Remote Sensing Image Semantic Segmentation

Semantic segmentation of remote sensing images plays an important role in many applications. However, a remote sensing image typically comprises a complex and heterogenous urban landscape with objects in various sizes and materials, which causes challenges to the task. In this work, a novel adaptive fusion network (AFNet) is proposed to improve the performance of very high resolution (VHR) remote sensing image segmentation. To coherently label size-varied ground objects from different categories, we design multilevel architecture with the scale-feature attention module (SFAM). By SFAM, at the location of small objects, low-level features from the shallow layers of convolutional neural network (CNN) are enhanced, whilst for large objects, high-level features from deep layers are enhanced. Thus, the features of size-varied objects could be preserved during fusing features from different levels, which helps to label size-varied objects. As for labeling the category with high intra-class difference and varied scales, the multiscale structure with a scale-layer attention module (SLAM) is utilized to learn representative features, where an adjacent score map refinement module (ACSR) is employed as the classifier. By SLAM, when fusing multiscale features, based on the interested objects scale, feature map from appropriate scale is given greater weights. With such a scale-aware strategy, the learned features can be more representative, which is helpful to distinguish objects for semantic segmentation. Besides, the performance is further improved by introducing several nonlinear layers to the ACSR. Extensive experiments conducted on two well-known public high-resolution remote sensing image data sets show the effectiveness of our proposed model. Code and predictions are available at https://github.com/athauna/AFNet/

Intelligent monitor for typhoon in IoT system of smart city

Accidents often occur in the earth—typhoons, floods, earthquakes, traffic accidents and so on. Whether these accidents can be timely and effectively responded to has been an important indicator to judge whether a region is advanced or not. IoT provide a possibility to solve such emergent problems by intelligent monitoring, diagnosis and repair. For example, coastal cities are often attacked by typhoons, if typhoon meteorological identification and early warning can be effectively carried out, many unnecessary property and personnel losses can be reduced. Accurate typhoon prediction has very important practical significance. However, current typhoon monitoring and prediction are mainly based on simulation with meteorological data; the accuracy still needs to be improved. Nowadays, the technology of Internet of Things (IoT) and remote sensing technology become more and more closely linked; many IoT systems in smart cities’ can obtain high-resolution remote sensing image data. Therefore, it is possible to use urban IoT system to realize the early warning of typhoon. In this paper, we propose a deep learning method for typhoon cloud recognition and typhoon center location, and design a general algorithm framework, including data preprocessing, model training and parameter selection, test and result analysis. Besides, we implement a typhoon early warning demo system. The experimental results show that our algorithm is better than the traditional methods in recognition accuracy.

A Theoretical Review of Vegetation Extraction Methods Based on UAV

At present, with the development of remote sensing technology and UAV technology, high-resolution remote sensing image has become one of the main data sources for vegetation extraction due to its rich texture features and geometric features. How to accurately and quickly extract ground object information from high-resolution images has become a hot topic in computer vision, photogrammetry and GIS research in recent years. At present, it is extremely convenient to acquire high-resolution remote sensing image data, while the extraction of ground object information, especially the vegetation classification and the research on information extraction methods, is relatively backward, which largely depends on the traditional classification methods. This paper summarizes the research progress of remote sensing classification of uav vegetation at home and abroad.

Color Difference Optimization Method for Multi-source Remote Sensing Image Processing

This paper proposes an approach to solve the problem of color inconsistency after the fusion and splicing of multi-spectral images of multi-source domestic satellites, by using regularly division the high-resolution remote sensing images, block by block computing the multiple linear regression and percent clip stretch. First of all, the remote sensing image is divided into blocks and multiple linear regression is carried out separately in each block. Secondly, on the basis of the image to be corrected after regression, percentage truncation and stretching are carried out on the over-concentrated gray band, so that the gray histogram distribution of the gray band is similar to that of the reference image. This scheme ensures that all pixels participate in the calculation, which not only improves the accuracy of the model, but also avoids the problems of long processing time and memory overflow caused by the large amount of high-resolution remote sensing image data. At the same time, each scene image selects an appropriate truncation percentage according to the evaluation index to solve the problem of excessive color difference caused by too concentrated image band gray. The experimental results show that the scheme in this paper can effectively solve the problems of too large color difference and too concentrated gray scale in image band. Compared with traditional methods, it has better performance in model accuracy, color retention and time efficiency.

RESEARCH ON HIGH RESOLUTION REMOTE SENSING IMAGE CLASSIFICATION BASED ON SEGNET SEMANTIC MODEL IMPROVED BY GENETIC ALGORITHM

Abstract. SegNet model is an improved model of Full Convolutional Networks (FCN). Its encoder, i.e. image feature extraction, is still a convolutional neural network (CNN). Aiming at the problem that most traditional CNN training uses error back propagation algorithm (BP algorithm), which has slow convergence speed and is easy to fall into local optimum solution, this paper takes SegNet as the research object, and proposes a method of extracting partial weights by using genetic algorithm (GA) to select features of SegNet model, and to alleviate the problem that SegNet is easy to fall into local optimal solution. In the training process of SegNet model, the weight of convolution layer of SegNet model used to extract features is optimized through selection, crossover and mutation of genetic algorithm, and then the improved SegNet semantic model (GA-SegNet model) is obtained by GA. In order to verify the image classification effect of the proposed GA-SegNet model, the same high-resolution remote sensing image data are used for experiments, and the model is compared with maximum likelihood (ML), support vector machine (SVM), traditional CNN and SegNet semantic model without GA improvement. The experimental results show that the proposed GA-SegNet model has the best classification accuracy and effect, which GA overcomes the problem of premature convergence of BP random gradient descent to a certain extent, and improves the classification performance of SegNet semantic model.

Lifelong Learning for Scene Recognition in Remote Sensing Images

The development of visual sensing technologies has made it possible to obtain some high resolution and to gather many high-resolution satellite images. To make the best use of these images, it is essential to be able to recognize and retrieve their intrinsic scene information. The problem of scene recognition in remote sensing images has recently aroused considerable interest, mainly due to the great success achieved by deep learning methods in generic image classification. Nevertheless, such methods usually require large amounts of labeled data. By contrast, remote sensing images are relatively scarce and expensive to obtain. Moreover, data sets from different aerospace research institutions exhibit large disparities. In order to address these problems, we propose a model based on a meta-learning method with the ability of learning a classifier from just few-shot samples. With the proposed model, the knowledge learned from one data set can be easily adapted to a new data set, which, in turn, would serve in the lifelong few-shot learning. Scene-level image recognition experiments, on public high-resolution remote sensing image data sets, validate our proposed lifelong few-shot learning model.

A multivariable optical remote sensing image feature discretization method applied to marine vessel targets recognition

The effective extraction of continuous features in ocean optical remote sensing image is the key to achieve the automatic detection and identification for marine vessel targets. Since many of the existing data mining algorithms can only deal with discrete attributes, it is necessary to transform the continuous features into discrete ones for adapting to these intelligent algorithms. However, most of the current discretization methods do not consider the mutual exclusion within the attribute set when selecting breakpoints, and cannot guarantee that the indiscernible relationship of information system is not destroyed. Obviously, they are not suitable for processing ocean optical remote sensing data with multiple features. Aiming at this problem, a multivariable optical remote sensing image feature discretization method applied to marine vessel targets recognition is presented in this paper. Firstly, the information equivalent model of remote sensing image is established based on the theories of information entropy and rough set. Secondly, the change extent of indiscernible relationship in the model before and after discretization is evaluated. Thirdly, multiple scans are executed for each band until the termination condition is satisfied for generating the optimal number of intervals. Finally, we carry out the simulation analysis of the high-resolution remote sensing image data collected near the coast of South China Sea. In addition, we also compare the proposed method with the current mainstream discretization algorithms. Experiments validate that the proposed method has better comprehensive performance in terms of interval number, data consistency, running time, prediction accuracy and recognition rate.

Impacts of land consolidation on rural human–environment system in typical watershed of the Loess Plateau and implications for rural development policy

Exploring the impact of land consolidation on the evolution of rural human-environment system (HES) is of positive significance for optimizing land consolidation model and innovating rural development policy. Taken a typical small watershed in Yan'an city as case study area, this paper explored the impact of gully land consolidation (GLC) on local HES from the perspectives of land use, landscape pattern, ecological security, social-economic response and comprehensive evaluation, based on high-resolution remote sensing image data, landscape pattern analysis and household surveys. The results showed that: (1) GLC could contribute to the improvement of land use structure. The terraces, sloping fields, shrub land and grassland at the bottom and both sides of the gully were mostly converted to high quality check dam land. Some of the shrub land were converted to more ecologically suitable native forest due to biological measures. (2) GLC could also help to optimize the landscape pattern. The average patch area and patch cohesion index of the check dam land increased, which indicated that the function of production improved. The landscape shape index and patch cohesion index of forestland and shrub land kept at a high level, and thus their ecological function was stable. At the watershed level, the fragmentation degree of landscape decreased, the landscape tended to be more diversified and balanced, and the anti-jamming capability of landscape and stability of ecosystem improved. (3) GLC have positive effects on the ecological security. Vegetation cover, ecological environment and capacity of flood control improved significantly, and soil erosion decreased by 55%. And (4) human activities responded to the changes of geographical environment. The scale of agricultural production, agricultural mechanization, diversity and non-agriculturalization of employment increased. However, the structure of agriculture is still unitary, the efficiency of agriculture is still low, and rural development is still relatively lagging. Field investigations showed that insufficient public participation, and lack of leadership and overall planning for sustainable utilization of resources and environment are the main reasons. A ladder model has been proposed for better understanding the impacts of land consolidation on territorial human-environment system in rural China. This paper suggests that land consolidation should be seen and evaluated in a broader rural development context, and more efforts should be made to innovate the rural governance mechanism of “common consultation and collective action”, thereby exerting the comprehensive effectiveness of land consolidation and promoting the transformation development and revitalization of territorial rural HES.

Land use and landscape change driven by gully land consolidation project: A case study of a typical watershed in the Loess Plateau

Exploring the impact of land consolidation on the changes of local land use and the landscape patterns is important for optimizing land consolidation models and thus accelerating the sustainable development of local communities. Using a typical small watershed in Yan’an City (Shaanxi, China), the impact of gully land consolidation on land use and landscape pattern change, based on high-resolution remote sensing image data and landscape pattern analysis, was investigated. The results showed that: (1) The terraces, sloping fields, shrub land and grassland at the bottom and both sides of the gully were converted mainly to high quality check dam land. Also, some of the shrub land, due to biological measures, was converted to more ecologically suitable native forest. Thus, the areas of check dam land and forests increased by 159 and 70 ha, while that of shrub land, grassland and sloping fields decreased by 112, 63 and 59 ha, respectively. (2) The average patch area and patch cohesion index for the check dam land increased, which indicated that the production function improved. The landscape shape index and the patch cohesion index for forestland and shrub land were maintained at a high level, and thus the ecological function remained stable. (3) At the watershed level, the degree of fragmentation of the landscape decreased and the landscape became more diversified and balanced; the anti-jamming capability of the landscape and the stability of the ecosystem improved also. Research suggests that implementing gully land consolidation in a rational manner may contribute to improvements in the structure of local land use and the patterns of landscape.

Measuring Community Greening Merging Multi-Source Geo-Data

Urban residential greening provides opportunities for social integration and physical exercise. These activities are beneficial to promoting citizens’ mental health, relieving stress, and reducing obesity and violent crimes. However, how to measure the distribution and spatial difference of green resources in urban residential areas have been controversial. This study takes the greening of urban residential units in Shenzhen City as its research object, measures the various greening index values of each residential unit, and analyses the spatial distribution characteristics of residential greening, regional differences, and influencing factors. A large sample of street view pictures, urban land use and high-resolution remote sensing image data are employed to establish an urban residential greening database containing 14,196 residential units. This study proposes three greening indicators, namely, green coverage index, green view index, and accessible public green land index, for measuring the green coverage of residential units, the visible greening of surrounding street space and the public green land around, respectively. Results show that (1) the greening level of residential units in Shenzhen City is generally high, with the three indicators averaging 32.7%, 30.5%, and 15.1%, respectively; (2) the types of residential greening differ per area; and (3) the level of residential greening is affected by development intensity, location, elevation and residential type. Such findings can serve as a reference for improving the greening level of residential units. This study argues that one indicator alone cannot measure the greenness of a residential community. It proposes an accessible public green land index as a measure for the spatial relationship between residential units and green lands. It suggests that future green space planning should pay more attention to the spatial distribution of green land, and introduce quantitative indicators to ensure sufficient green lands around the walking range of residential areas.

U-net Network for Building Information Extraction of Remote-Sensing Imagery

With the resolution of remote sensing images is getting higher and higher, high-resolution remote sensing images are widely used in many areas. Among them, image information extraction is one of the basic applications of remote sensing images. In the face of massive high-resolution remote sensing image data, the traditional method of target recognition is difficult to cope with. Therefore, this paper proposes a remote sensing image extraction based on U-net network. Firstly, the U-net semantic segmentation network is used to train the training set, and the validation set is used to verify the training set at the same time, and finally the test set is used for testing. The experimental results show that U-net can be applied to the extraction of buildings.

An Approach for Estimating Solar Photovoltaic Potential Based on Rooftop Retrieval from Remote Sensing Images

Solar energy is the most clean renewable energy source and has good prospects for future sustainable development. Installation of solar photovoltaic (PV) systems on building rooftops has been the most widely applied method for using solar energy resources. In this study, we developed an approach to simulate the monthly and annual solar radiation on rooftops at an hourly time step to estimate the solar PV potential, based on rooftop feature retrieval from remote sensing images. The rooftop features included 2D rooftop outlines and 3D rooftop parameters retrieved from high-resolution remote sensing image data (obtained from Google Maps) and digital surface model (DSM, generated from the Pleiades satellite), respectively. We developed the building features calculation method for five rooftop types: flat rooftops, shed rooftops, hipped rooftops, gable rooftops and mansard rooftops. The parameters of the PV modules derived from the building features were then combined with solar radiation data to evaluate solar photovoltaic potential. The proposed method was applied in the Chao Yang District of Beijing, China. The results were that the number of rooftops available for PV systems was 743, the available rooftop area was 678,805 m2, and the annual PV electricity potential was 63.78 GWh/year in the study area, which has great solar PV potential. The method to perform precise calculation of specific rooftop solar PV potential developed in this study will guide the formulation of energy policy for solar PV in the future.

CLUSTERING OF HIGH RESOLUTION UAV IMAGERY TO IDENTIFY ESSENTIAL PLANTS USING SOM NEURAL NETWORK

The use of high-resolution remote sensing image data is necessary to distinguish essential plants with other plants. This study uses image data taken using Unmanned Aerial Vehicle (UAV) to identify essential plants especially citronella and kaffir lime. To distinguish the structure of essential plants with other objects used texture features extracted by wavelet daubechies method. The features that have been ekstract, then is grouped based on the proximity feature with the Self Organizing Map (SOM) Neural Network. Thus, objects that have similar features will clump together. The tests were conducted on two groups of data sets, where the first group data consisted of plants, buildings and vacant lots, while the second group data consisted only of plants. The results of testing of the first data group shows that the techniques can recognize the citronella plants among other objects, especially building objects and bare land with purity of 0.862745 and Silhouette Coeficient of 0.5520671. While in the second data group, the value of purity and Silhouette Coeficient decreased to 0.737705 and 0.161028. However, from the test of the second data group still shows that the method used can distinguish citronella crops to other plants.

基于形态学纹理特征的高分辨率遥感影像变化检测

由于高分辨率遥感影像数据可以提供更加精细和准确的地表类别信息，它已经成为土地利用覆盖变化研究的主要数据源之一。本文以15米空间分辨率的遥感影像为数据源，利用数学形态学运算得到了高分辨率遥感影像的纹理特征图像序列，进行了决策树分类，得到了土地利用变化前后期的图像类别信息，采用图像变化轨迹法来进行变化检测。实验结果证明了本文提出算法的有效性，为更加实用的变化检测研究打下了基础。 Because high-resolution remote sensing image data can provide more detailed and accurate surface classification information, it has become one of the main data sources of land use change research. In this paper, the remote sensing image with spatial resolution of 15 m is used as the data source, and the texture feature sequence of high resolution remote sensing image is obtained by mathematical morphology. The decision tree classification is made, and the image classification information is obtained before and after land use change. Land use change analysis was carried out by comparative method. The experimental results show that the proposed algorithm is effective and provides the basis for more practical research on the change detection.

Visual attention model based mining area recognition on massive high-resolution remote sensing images

With the development of remote sensing technology, satellite images with the characteristics of multi-scale, multi-band, and multi-date make it tend to be big data. So how to raise the extraction speed, precision and automatic degree of salient objects from high-resolution remote sensing images become urgent problems. Based on the analysis using an Itti visual attention model for natural image processing, we achieved improvements in two aspects: (1) the selection of salient regions based on elevation data, and (2) the segmentation of salient regions using the Snake model for precise object contour extraction. Tests on the extraction of 2.5 m high-resolution remote sensing image data in the rare earth mining area in Dingnan County, Jiangxi Province showed a false alarm rate of 14.8 % and a missing alarm rate of 8.4 % in the extraction of mine quantity data. The proposed method could be useful for improving the speed, precision and automatic extraction of salient objects from high-resolution remote sensing images as well as the boundary information of salient objects that are based on a visual attention model.