最新刊期
卷 25 ， 期 3 ， 2021
-
摘要：With the rapid development of aerospace technology and the continuous increase of China’s high resolution earth observation data acquisition in the past few decades, the era of remote sensing big data has coming now. Developing multi-satellite integrated data processing and application technology has become an important trend. In this paper, we systematically review the technological development process from two aspects, including multi-satellite imaging processing and multi-element information extraction. Then, we analyze the advantages and characteristics of the existing cutting-edge methods, and points out the main challenge of establishing the integrated processing model for imaging processing field and efficient learning model for information extraction field. On this basis, as well as according to the practical application requirements, a novel method of multi-satellite integrated processing and analysis under remote sensing big data is proposed in this paper. We emphatically define the basic concepts, scientific problems, research ideas and solutions of this method. On one hand, in terms of multi-satellite imaging processing, aiming at the problem of difficult estimation of high-dimensional coupled imaging parameters at high resolution, the various errors caused by the whole cycles of electromagnetic waves are calculated quantitatively, including payload error, platform error, data transmission error, atmospheric influence and so on. Then a multi-satellite integrated imaging processing physical model based on the Generative Adversarial Networks (GAN) is constructed to approximate estimate imaging parameters. In this way, we can achieve high-precision geometric corrections and radiation corrections, as well as generate high-quality remote sensing image products. On the other hand, in terms of multi-element information extraction, to solve the problem that the accuracy of the original tasks is difficult to maintain due to the addition of new tasks and requires full sample retraining, we design a multi-task feature sharing model based on few-shot incremental learning. It has a novel memory retention unit and multi-modal joint optimization of convex non-negative matrix factors. Through this model, we can generate parallel and high-precision annotations for multiple objects. Compared with the existing methods, the information between different satellites and payloads, different missions and objects are complementary to each other, leading to the simultaneous improvement of multi-sensor imaging quality and object extraction accuracy. The specific technical approach and preliminary experimental verification are given in detail in this paper. Experiments on multi-satellite imaging processing show that our method can effectively estimate multiple imaging errors. The phase estimation accuracy is within 1 degree when the signal-to-noise ratio is -5dB, which indicates the good performance even under low signal-to-noise ratio condition. At the same time, experiments on multi-element information extraction show that compared to single-modal method, our method, which uses multi-modal data in combination and embedded the novel memory retention unit, improve the extraction accuracy in multi-tasks of object detection and semantic segmentation. In the future, we will pay more attention to the disconnection problem between multi-satellite imaging processing and multi-element information extraction in the field of remote sensing. By establishing a benign mutual feedback mechanism between these two procedures to maximize the benefits of the remote sensing big data.关键词：remote sensing big data;multi-satellite integrated;multi-satellite imaging processing;multi-element information analysis;Generative Adversarial Networks(GAN);multi-task- 803
- |
- 264
- |
- 6
发布时间：2021-03-29
Research Progress
-
摘要：Geospatial information is important in the era of artificial intelligence and big data. Small, lightweight unmanned aerial vehicles (UAVs) for aerial photogrammetry and Remote Sensing (RS) technology, as the main means of obtaining centimeter-scale resolution and real-time remote sensing data, may be expected to play an important role in these fields.First, this paper focuses on the development status and trends of UAV aerial photogrammetry and RS systems. Fixed-wing, lightweight, and small UAVs are an early type of aircraft in the field of surveying and mapping. The hand-throwing and vertical take-off and landing characteristics of fixed-wing UAV systems have prompted the development of these technologies toward the intelligent direction. Multi-rotor UAVs are important instruments in remote sensing mapping, but their flight duration requires further improvement. Unmanned helicopters are widely used in the remote sensing mapping of heavy loads, but this technology is greatly affected by cost, control complexity, and other factors. Digital camera, video camera, tilt camera, lidar, SAR, POS, and other loads are highly useful in surveying and mapping; however, because the detection and calibration of new sensors and UAV systems has not been optimized, the development of new technologies and equipment is limited to a certain extent.Next, the problems and challenges of system detection, large-scale real-time remote sensing, and accurate big data interpretation of UAV aerial photogrammetry and RS data are summarized and analyzed. UAV aerial photogrammetry and RS systems are widely used, but the detection methods and standards of UAV systems aimed at environmental adaptability, flight performance, navigation control accuracy, and electromagnetic compatibility have not been perfected. The real-time multi-level transmission technology of video, imagery, and other surveying and mapping data must be further developed for emergency rescue. The data obtained each year by UAV aerial photogrammetry and RS are more than PB and mostly used to produce standard surveying and mapping products. However, the data collected are insufficient for data mining and analysis.Finally, facing the technical backgrounds of artificial intelligence, big data, Internet of things, and cloud computing, among others, future development directions of intelligent flight control, UAV operation, and real-time, real-scene UAV remote sensing technologies are provided. The continuous development and improvement of the industry and rapid promotion of market demands, aerial photogrammetry, and RS technology of lightweight and small UAVs may be expected to promote industrial changes in earth observation and accurate public perception, potentially forming a new industry worth over 100 billion US dollars.关键词：light-weighted and small UAV for aerial photogrammetry and RS;aerial photogrammetry and RS sensors;artificial intelligence;big data- 1361
- |
- 626
- |
- 18
发布时间：2021-03-29 -
摘要：Mangrove forest provides huge value of ecosystem services, such as beach protection, siltation promotion, flood and wave prevention, preventing waves, and biodiversity maintenance. Species composition and diversity are key parameters for assessing the health of forest ecosystem, and the loss of species diversity often accelerates degradation of structure and function of forest ecosystem. Therefore, accurately monitoring species composition and spatio-temporal distribution of mangrove forest are helpful for timely and effective management and restoration measures, which can further retain the quality of the mangrove ecosystem and biodiversity. The traditional means of obtaining mangrove species information requires time-consuming, labor-intensive and costly field survey, however, it is difficult to further understand the continuous distribution of forest health. In contrast, remote sensing technology is more cost-effective and can achieve spatially continuous monitoring of mangrove species composition and health status. With the fast-developing fine-resolution multispectral satellites, the images are used for classifying mangrove species due to their rich spatial geometric information. However, compared to multispectral images that contain limited spectral information, hyperspectral data are more effective for tree species discrimination or classification due to hundreds or even thousands of continuous bands that can reflect vegetation functional traits (e.g. pigment content, specific leaf area and nitrogen content). Moreover, LiDAR (Light Detection and Ranging) point cloud can acquire details related to three-dimensional features of the vegetation structure.Traditionally, there are many data dimension reduction or feature extraction methods, such as Principal Component Analysis (PCA), Independent Component Analysis (ICA), Minimum Noise Fraction (MNF) and Successive Projections Algorithm (SPA). However, these methods should combine with other classifiers in classification of plant species. To improve the classification accuracy and efficiency, this paper introduces a new machine learning method with both feature selection and classification functions—eXtreme Gradient Boosting (XGBoost). The rapidly growing Unmanned Aerial Vehicles (UAV) and portable sensors of hyperspectral imagery and LiDAR have provided higher quality remote sensing data. UAV-based remote sensing data can derive massive features of spectra, texture and structure, therefore, how to extract dominant features is a key issue to improve the efficiency and accuracy of mangrove species classification.With UAV-based hyperspectral imagery and LiDAR point cloud of the buffer area in Shenzhen Futian Mangrove Nature Reserve, this study aims to extract eight types of dominant features suitable for mangrove species classification using “feature_importance” property of XGBoost. The dominant features include: single feature derived from UAV-based hyperspectral imagery (spectral bands, vegetation indices and texture: F1—F3) and their fused feature (F4), single feature derived from UAV-based LiDAR point cloud (height and intensity feature: F5, F6) and their fused feature (F7), and fused feature coupling hyperspectral imagery and LiDAR point cloud.Synthetically considering species classification accuracy and mapping results, the classification model based on F8 feature held the best performance (overall accuracy was 96.41%, Moran’s I was 0.5520). The classification performance based on fused feature of single data source (F4 and F7, overall accuracy was separately 96.74% and 90.64%) was superior than that of single feature (F1—F3, F5 and F6, overall accuracy was 90.31%, 92.20%, 91.96%, 87.66% and 81.99% respectively). The Moran’s I of classification maps based on fused feature (F4, F7 and F8) and texture feature (F3) were greater than that of single feature (F1, F2, F5 and F6). Moreover, mangrove species classification models based on different dominant features have their own advantages on spatial mapping. The introduction of F3 effectively solved the common salt-and-pepper effect in the mapping results based on F1 and F2; moreover, the salt-and-pepper effect in the edge of classification images (near tidal flat area) was significantly improved in the mapping results based on F5, F7 and F8.We conclude that: (1) The combination of UAV-based remote sensing data and XGBoost is feasible to pixel-oriented accurate classification of mangrove species, the fusion feature of UAV-based hyperspectral image and LiDAR point cloud has the best classification effect when comprehensively comparing classification accuracy (OA) and mapping effect (Moran’s I); (2) when fusing the different types of features derived from UAV-based hyperspectral or LiDAR data alone, the corresponding classification accuracy and mapping effect behaved better than the single feature derived from the same UAV data source; (3) XGBoost, a machine learning method with both feature selection and classification functions, has great potential in remote sensing image classification; (4) the intensity features derived from LiDAR point clouds are greatly affected by UAV flight strips, however, the height features are robust in mangrove species classification. Future research will focus on unmixing of hyperspectral data, fuzzy classification and radiation transfer models (such as PROSAIL) to improve the accuracy and interpretability of mangrove species classification. This paper demonstrated the feasibility of UAV-based remote sensing data and XGBoost in the pixel-oriented precise classification of mangrove species, which can provide scientific basis and technical support for three-dimensional monitoring of health, conservation and restoration for mangrove ecosystem.关键词：remote sensing;mangrove;Tree Species Classification;UAV;hyperspectral imagery;LiDAR point cloud;XG1300st- 1057
- |
- 371
- |
- 19
发布时间：2021-03-29
UAV Remote Sensing
-
摘要：Sea ice concentration, which refers to the percentage of sea ice in an area, is an important parameter describing the characteristics of sea ice. Remote sensing monitoring of sea ice is crucial to understand the role of polar regions in the global climate system and global warming. Sea ice information is of great importance in ship transportation, weather forecasting, and global climate forecast. At present, passive microwave radiometers are the main means of monitoring sea ice concentration; however, because microwaves present wavelength limitations, conventional sea ice concentration products cannot be used in practical applications in small areas. Visible-light-infrared remote sensing can retrieve sea ice concentrations, and its advantages over other sensing methods include high spatial resolution.This work focuses on the development of a sea ice concentration algorithm suitable for medium-resolution images. Few medium-resolution sea ice concentration products based on visible-light remote sensing are available, and only NOAA has released relevant operational products. However, the accuracy of its algorithm for low-concentration sea ice inversion is low. This paper proposes an improved algorithm based on the existing algorithm to determine the ice node via the nearest-pixel method. MODIS data are used as a data source to calculate the sea ice concentration, and Landsat 8 OLI data with a spatial resolution of 30 m are used for comparative verification.Results show that the improved algorithm can improve the inversion accuracy of low-concentration sea ice. The Liu algorithm has the disadvantage of overestimation. In the case of sea ice concentrations of 0% — 50%, the average deviation of the Liu algorithm is 13%, and its standard deviation is 38%. By comparison, the average deviation of the improved algorithm is 5%, and its standard deviation is 32%. In the case of sea ice concentrations of 0% — 100%, the average deviation of the Liu algorithm is 4%, and its standard deviation is 32%. By comparison, the average deviation of the improved algorithm is -3%, and its standard deviation is 28%. In the case of sea ice concentrations of 0% —50%, the accuracy of the improved algorithm is better than that of the Liu algorithm. When the sea ice concentration is close to 100%, the results of the two algorithms are highly similar. Overall, the improvement effect of the proposed algorithm is related to the actual sea ice concentration, and the improved algorithm is more accurate than the Liu algorithm for low-concentration sea ice regions, such as ice-water transitions and broken ice coverage.关键词：remote sensing;sea ice;ice water recognition;MODIS;nearest pixel;SIC algorithm- 713
- |
- 309
- |
- 3
发布时间：2021-03-29 -
摘要：The low efficiency of orbit data processing and poor effectiveness of remote sensing satellite information acquisition are problems requiring great attention. Meeting the requirements of satellites in orbit in real time and efficient extraction of the targets of interest are challenging endeavors because of their heavy dependence on data transmission and communication bandwidth.In this paper, an efficient ship target detection algorithm based on global statistics and an ultra-lightweight suspected target identification network is designed to achieve the rapid extraction of ship targets. The traditional image processing method is used to detect the target quickly, obtain the suspected target slice, and reduce the amount of data significantly. A self-designed and improved ultra-lightweight identification network based on deep learning is then used to realize a second cycle of suspected target screening and improve the accuracy of target extraction. Much work has been done to simplify the real-time implementation of image-processing methods, such as OTSU threshold calculation, connected domain labeling, and target identification network based on deep learning. The real-time processing accuracy, speed, hardware scale, and heat consumption of the proposed method are well balanced by reasonable optimization of the algorithm flow and calculation method and establishment of an appropriate error analysis model. The calculation cost of the algorithm is low, and the dependence of the real-time implementation of the algorithm on the performance of the hardware platform is reduced greatly, thereby ensuring the excellent overall performance of the algorithm. In particular, network pruning, weight parameter sharing, and quantization are used to reduce the network weight parameter and forward reasoning calculation storage requirements, thus improving the design of the target identification network.GF-3 satellite data were used to test the algorithm, and experimental results showed that the accuracy of ship target extraction could be improved by 20%—98%, the computational complexity could be reduced by 90%, and the real-time performance could be improved by 50%. When the ship target extraction algorithm model was implemented in a low-power embedded circuit, the power consumption of the whole calculation circuit was less than 13 W, and the standby power consumption was less than 1.5 W; these values meet the requirements of full-time operations in orbit.The proposed method takes into account the effectiveness of the algorithm and feasibility of in-orbit real-time processing to improve the energy efficiency ratio and target extraction performance of the system effectively. The proposed algorithm was implemented in the current satellite embedded circuit and the orbit of a new radar test satellite, and the results obtained reflect good application prospects.关键词：ultra-lightweight network;SAR image;target detection;target identify;real time processing on orbit- 929
- |
- 303
- |
- 5
发布时间：2021-03-29 -
摘要：Improving the spatial resolution of microwave Soil Moisture (SM) production is of great significance for hydrological and agricultural applications on a regional scale. Downscaling microwave satellite SM with optical/thermal infrared and microwave fusion method shows great application potential. However, it mostly relies on remote sensing surface temperature (LST) or the SM index derived by LST decomposition, which is limited by the cloud contamination problems, LST decomposition uncertainties, and the decoupling effect between LST and SM. To circumvent these problems, we made a primary study on downscaling microwave SM by coupling MOD16 and SMAP data. In this study, we constructed three parameterized downscaling functions (i.e., exponent, cosine, cosine squared) between Land surface Evapotranspiration Efficiency (LEE) and SM. MOD16 products is employed to calculate LEE, which has a spatial resolution of 500 m. Combining the parameterized downscaling functions and the high-resolution LEE, original SMAP SM (spatial resolution, 36 km) data were successfully downscaled to a spatial resolution of 500m. The downscaled SM was evaluated in terms of dynamic range, energy conservation, in situ SM at sparse stations, and in situ SM at Core Validation Station (CVS). Results demonstrated that the downscaling algorithm increases the spatial detail characteristics of original SM, maintains the dynamic range of SM, and preserves energy during the downscaling process. Moreover, it maintains the performance of the original SM as compared with in situ SM at CVS and sparse stations. Sensitivity analysis showed that the cosine-square downscaling function is less sensitive to errors in MOD16 production than the other two downscaling functions.关键词：microwave soil moisture;spatially downscaling;land surface evapotranspiration efficiency;MOD16;SMAP- 724
- |
- 221
- |
- 2
发布时间：2021-03-29 -
摘要：Remote sensing images with both high spatial and high temporal resolutions are highly desirable in various applications. However, due to technical limitations of remote sensing imaging system and other factors, the acquired remote sensing images have to make a fundamental trade-off between high spatial and temporal resolutions. For example, the MODIS images have high temporal resolution, its spatial resolution is low; On the contrary, the Landsat images have high spatial resolution with relatively lower temporal resolution.Spatio-temporal fusion can integrate the complementary advantages of high spatial resolution and high spectral resolution, respectively, of multi-source remote sensing images, to generate time-continuous images with high spatial resolution. This has important application value in remote sensing image dynamic monitoring, time-series analysis, and other aspects. To the best of our knowledge, at present, most of studies generally evaluated the spatio-temporal fused images based on a single type of remote sensing data, such as the surface reflectance data or the Normalized Difference Vegetation Index (NDVI) remote sensing product, etc. However, how well a spatio-temporal fusion method performs in practical applications? This should be comprehensively assessed from different aspects based on different types of remote sensing data products. In addition, most of studies performed the evaluation of a spatio-temporal fusion method based on the fused image at a single phase. However, for spatio-temporal fusion, the final target is actually to obtain time-series fused images, the quality evaluation of the fused images from the temporal dimension should be also taken into account. Whereas, to the best of our knowledge, the quality evaluation for time-series fused images is not comprehensively considered in the existing studies. In this paper, we proposed to evaluate the spatio-temporal fusion methods from the comprehensive perspective of single time point, time series, and multiple different remote sensing data products. In this paper, spatio-temporal fusion data sets, including surface reflectance data set, NDVI data set, and the Land Surface Temperature (LST) data set were established based on Landsat and MODIS remote sensing satellite images. In addition, some typical spatio-temporal fusion methods were reviewed, and the performance of four spatio-temporal fusion algorithms, including the Spatial and Temporal Adaptive Reflectance Fusion Model (STARFM), Enhanced STARFM (ESTARFM), Flexible Spatio-Temporal Data Fusion (FSDAF), and the Spatial and Temporal Nonlocal Filter-based Fusion Model (STNLFFM), were qualitatively and quantitatively evaluated based on the proposed data sets of different kinds of remote sensing data products, i.e., the surface reflectance, NDVI, and LST. In addition, the quality evaluation from the perspective of both single-time-point and time-series dimensions were performed. The experimental results show that the performance of spatio-temporal fusion algorithms can be more comprehensively verified based on different type of data sets, and the evaluation combined with single time point and time-series data set is more objective.关键词：spatio-temporal fusion;reflectance;Normalized Difference Vegetation Index(NDVI);Land Surface Temperature(LST);quality evaluation- 720
- |
- 251
- |
- 0
发布时间：2021-03-29 -
摘要：The Visible and Infra-Red Radiometer (VIRR) sensor on FY-3C was affected by the scanning mirror of the remote sensing instrument illuminated by the sunlight at high latitude, which results in the noise of the earth observation image, and the stripe noise of the channel-3 in VIRR, which seriously affected the data application. The causes of strip noise pollution include the influence of direct sunlight near the terminator, the effect of space clamp caused by stray light formed by reflection and scattering of sunlight and the influence of temperature fluctuation of scanning mirror caused by sunlight.According to the anisotropic characteristics of the stripe noise and the unidirectional variational model, We studied the removal of stripe noise in channel 3 of VIRR, and compared the results with the low-pass filtering method and TV-L1 method. The mean cross-track profiles before and after destriping, Peak Signal to Noise Ratio (PSNR), Improvement Factors (IF) of radiation quality and Inverse Coefficient of Variation (ICV) were used to evaluate the destriping results. In addition, in order to analyze the change of solar pollution over time, we made pollution line statistics, using the data of January, April, July and October 2014-2019.The results show that the variational model had a good effect on the stripe noise caused by solar pollution in the observation data of FY-3C VIRR channel 3. In the real experiment, PSNR was increased to 32.77 db; in the real data experiment, IF was increased to 16.99 db. The results of time series analysis of solar pollution show that solar pollution has significant seasonal variation and has significant correlation with satellite β angle.关键词：visible and infra-red radiometer(VIRR);solar pollution;variational model;destriping;FY-3C- 627
- |
- 291
- |
- 2
发布时间：2021-03-29
Technologies and Methodologies
-
摘要：Wind damage is one of the most serious natural disasters affecting the development of China’s natural rubber industry. It caused huge physical damage (e.g., a large number of fallen leaves, branches and trunk breakage) to rubber plantation in a short period of time, which seriously affected the subsequent growth and latex production. Although traditional ground surveys have high accuracy, they are time-consuming, labor-intensive, and economical. The rapid assessment using remote sensing is of great significance for guiding post-disaster production recovery, insurance compensation, and scientific research.Taking the tornado induced by tropical storm Yangliu in 2019 in western Hainan Island as a case study, this study explores the potential of combining Landsat and Sentinel-2 time series images to assess tornado damage of rubber plantations from the perspectives of data availability, image composite methods, and disaster assessment indicators. The image difference method was used to detect differences before and after the tornado.The results showed that: (1) cloudless Landsat 7/8 and Sentinel-2 images can cover more than 90% of the study area at least once within 20 days before and after the disaster, and almost cover the whole area within 30 days. The average coverage at the pixel scale is three times in 30 days and six times in 60 days. (2) The image difference generated by the maximum value composite of images acquired before tornado and the medium value composite of image acquired after tornado is the most significant and stable. (3) In term of time window, the monitoring effect from images acquired within 40 days before and after tornado tends to be stable, and there is no significant difference when compared with the results generated from a 90-days time window. It is conservatively recommended to use images in a 60-day window before and after the tornado for disaster assessment. (4) The EVI of the damaged rubber plantation before and after the tornado changed the most, followed by NBR, LSWI, NDVI, near infrared (NIR), and shortwave infrared (SWIR1 and SWIR2) bands. In terms of percentage changes, LSWI and SWIR2 showed the highest changes after the tornado, followed by NBR, EVI, NIR, and SWIR1. However, the spatial variation of LSWI was significantly lower than that of SWIR2. (5) The tornado completely destroyed about 645 ha of rubber plantation in western Hainan Island. As a crop with a 30-years economic life cycle, the loss is very serious.This study demonstrates the great potential of combining Landsat and Sentinel-2 images to access tornado damage of rubber plantation, and provides important insights for timely evaluating typhoon disaster of rubber plantations and other crops in future.关键词：rubber plantation;long time series images;image composite method;time window;damage assessment index- 859
- |
- 314
- |
- 4
发布时间：2021-03-29 -
摘要：The Arctic River Lena is the 10th longest river in the world and the second largest river in the Arctic region. The annual river discharge of the Arctic River Lena accounts for approximately 20% of the total freshwater in the Arctic Ocean. It also drains a large amount of organic matter from terrestrial ecosystems into the ocean and plays a very important role in the global carbon cycle. Satellite remote sensing data are considered a necessary supplement to the ground-based monitoring of riverine organic matter circulation, especially in high-latitude regions during the ice-free period.The objectives of this study are to (1) construct a high-accuracy retrieval algorithm to estimate the Dissolved Organic Carbon （DOC） concentration of the Arctic River Lena, (2) analyze the variation characteristics of DOC concentration over a long time series using remote sensing images, and (3) discuss the main driving factors of DOC concentration variation in the Arctic River Lena.In this paper, a remote sensing retrieval algorithm based on the Google Earth engine was constructed. Landsat images retrieved from 1999 to 2018 were used to obtain the concentration of Chromophoric Dissolved Organic Matter (CDOM) in the Arctic River Lena. Given the strong correlation between the field measurements of CDOM and dissolved organic carbon (R2 = 0.873), the CDOM retrieval results were converted to DOC concentrations in this paper. Thus, this paper analyzes the temporal and spatial dynamics of DOC in the Arctic River Lena during the ice-free period over the last two decades.Results showed that the performance of the retrieval algorithm supports the feasibility of using Landsat data of different sensors to monitor riverine DOC variations. The boosted regression tree model was used to analyze the doc variation of the Artic Lena River, which is influenced by many driving factors, including land cover change, watershed slope, meteorological factors, human activities, and latitudinal zonation. The seasonality, geography, and scale could affect quantitative relationships between DOC concentration and these influencing factors.In conclusion, our results could improve the ability to monitor DOC fluxes in Arctic rivers and advance our understanding of the Arctic’s carbon cycle.关键词：DOC;CDOM;Arctic River Lena;Landsat;long-term;Google Earth Engine;Boosted Regression Tree (BRT)- 648
- |
- 220
- |
- 2
发布时间：2021-03-29
Remote Sensing Applications
- Postal code：100190
- Tel：010-58887052 Email：nrsb@aircas.ac.cn
- Technical support is provided by Beijing Founder electronics co., LTD 京ICP备20021838号-9
京公网安备11010802024621 - It is recommended to read the content of this site in Chrome&IE9+. Please switch to extreme mode in browser 360.
- Cookies We use cookies to help provide and enhance our service and tailor content. By continuing, you agree to the use of cookies.
0