Skip to main content
main-content

Über dieses Buch

This book is the proceedings of the 7th China High-resolution Earth Observation Conference (CHREOC). The series conference of China High Resolution Earth Observation has become an influential academic event in the earth detection area, attracting more and more top experts and industry users of related fields. The CHREOCs focus on the popular topics including military-civilian integration, the One Belt and One Road project, the transformation of scientific research achievements. They also discuss the new ideas, new technologies, new methods, and new developments. The CHREOCs have effectively promoted high-level institutional mechanisms, technological innovation, and industrial upgrading in the high-resolution earth observation area, and extend the influences of the state-sponsored major projects.

Inhaltsverzeichnis

Frontmatter

Weight Factor Determination of Reverse Distance Weighting Method in Computation of Geomagnetic Diurnal Variation Data

A virtual geomagnetic diurnal station has been used to calculate the diurnal variation correction data in the regions where the ground magnetic surveying is difficult to be implemented, and this will provide diurnal variation correction data for the airborne geomagnetic measurement. The weight factors of reverse distance weighting (RDW) method usually use the empirical values, which have limitation in the application of different testing areas. The correlation coefficients are proposed and introduced to calculate the weight factors iteratively, which are more practical. The numerical experimental results show that the diurnal variation correction data, whose weight factors are calculated by correlation coefficients, have a higher precision. The precision of northern component X and horizontal component H are superior to 2.2 nT, eastern component Y and geomagnetic total intensity F are superior to 1.0 nT, vertical component Z is superior to 0.3 nT, and geomagnetic declination D and geomagnetic inclination I are superior to 10″. Therefore, the diurnal variation correction data in higher precision can be provided for the airborne scalar and vector geomagnetic measurement in the regions where the ground magnetic surveying is difficult.

Liu Xiaogang, Liu Qian, Wang Yunpeng, Song Ying

Optimal Rescheduling for OOS Mission Based on Squeaky-Wheel Optimization

On-Orbit Servicing (OOS) is the process of improving a space-based capability through a combination of in-orbit activities. With the successful completion of the OOS enabling technologies, the mission planning problem is issued, optimizing mission sequences and trajectories for Servicing Spacecraft (SSc) servicing multiple targets. During the mission process, even though the adequate preparations were done, some accidents may still occur, e.g. the user adds some unassigned targets in emergency, or one of the SScs malfunctions and the un-completed targets should be re-allocated. Then it is necessary to reschedule the existing mission scheme, regenerating a new mission scheme to deal with the new-arrival targets, minimizing the disturbances on the initial mission scheme, minimizing the total rendezvous cost increment, and at the same time designing the rendezvous trajectories. The OOS mission rescheduling problem for coplanar circular constellation is studied in this paper. The reschedule model is developed first. Then base on the Squeaky-Wheel Optimization framework, a general optimization framework for combinatorial optimization problems, the solution method is proposed. Numerical simulations are carried out to demonstrate the effectiveness of the model and rescheduling method.

Jing Yu, Dong Hao, Xiaoqian Chen

Method of the Stability of Soil Landslide Under the Condition of Continuous Rainfall

In recent years, landslide disasters occur frequently, making the evaluation of the susceptibility of landslide hazards a major difficulty and hot topic in current research. At present, the research mainly focuses on using statistical model or information model to analyze landslide disaster, but the accuracy is not high. In this study, the combination of Limit Equilibrium Slice Method and Finite Element Method is used to model the landslide and analyze the stability of soil landslide under complex conditions. On this basis, taking a typical soil landslide in a village of Shaanxi Province as the research area, combined with multi-source geological data, the possible landslide mass and slip surface under the condition of continuous rainfall are obtained, which provides effective technical support for landslide disaster prediction and prevention planning.

Min Yang, Yihan Xie

Processing and Application of AMS-3000 Large Field of View Triple Line Array Camera in Photogrammetry

AMS-3000 is a three-line stereo aerial camera with large field of view and high resolution. It can not only acquire panchromatic and RGB high-resolution images, but also has high accuracy. Compared with the world’s most advanced ADS40/80/100 camera, this type of camera has higher resolution and higher working efficiency. This article mainly analyzes the imaging principle, working principle, main technical parameters and working efficiency of AMS-3000. By comparing the focal length, the resolution, and coverage width at the same resolution with ADS40/80/100, the superiority of AMS-3000 has emerged. Then the processing flow of the AMS-3000 system is proposed. Next, take Yangjiang in China as the survey area, the good working efficiency and high imaging performance have been confirmed. Finally, the future development and prospect of the AMS-3000 camera are forecasted.

Shuping Xiong, Yansong Duan, Guoqin Yuan

Evaluation and Analysis of Geometric Accuracy of the ZY-1 02D VNIC Image

The ZY-1 02D reached orbit in September 12, 2019. It is equipped with a 9-band visible near-infrared camera (VNIC) and a 166-band hyperspectral camera. The VNIC can acquire a panchromatic/multi-spectral image with a spatial resolution of 2.5/10 m and a width of 115 km. The geometric accuracy of ZY-1 02D VNIC is evaluated using multiple revisited data in Northeast and Northwest China. Based on the ground control points (GCPs) extracted from the ZY-3 standard stereo triplet products, the accuracy of the bundle block adjustment with the affine correction in image space is analyzed, and the root mean square error (RMSE) of the panchromatic image is better than 0.5 pixel in most cases. However, through the static and dynamic residual analysis of GCP, it is found that the vibration of the satellite platform will cause significant dynamic imaging errors, especially in the push-broom direction along the orbit, an effect of approximately 1.0 pixels in size can be detected. In general, the panchromatic images of ZY-1 02D VNIC have good geometric accuracy and meet the requirements of design specifications. However, the accuracy of panchromatic image is adversely affected by platform vibration.

Liping Zhao, Xingke Fu, Xianhui Dou

Band Registration Analysis of the ZY-1 02D VNIC Image

The ZY-1 02D mission is mainly devoted to the operational services of monitoring, understanding, and managing natural resources in China. The Visible Near-Infrared Camera (VNIC) carried on the ZY-1 02D is equipped with 8 multispectral bands at 10 m resolution with a 115 km swath width. The registration accuracy of the multi-spectral bands is analyzed using the multiple images cloud free from northern China. For each image, the relative registration errors among all bands are measured based on more than 10,000 evenly distributed and highly accurate tie-points extracted using the phase correlation method. According to the design characteristics of VNIC, the 8 bands are divided into 2 groups, the first 4 bands are B1234, and the last 4 bands are B5678. In most cases, the root mean square error (RMSE) of the registration is up to 0.3 pixels within group B1234 or B5678. However, between B1234 and B5678, the registration accuracy drops to about 0.4 pixels. After preliminary analysis, this degradation is believed to be caused by the slight vibration of the satellite platform and the geometric distortion between the bands.

Liping Zhao, Hongzhou Li, Huai Wang

Analysis and Optimization of Geometric Accuracy of ZY-1 02C HRC Image

The ZY-1 02C is China’s first operational high-resolution civilian remote sensing satellite. It was launched on December 22, 2011, it has been in a continuous operation on orbit for more than 9 years. It is equipped with a panchromatic and multispectral scanner and a panchromatic High Resolution Camera (HRC) with 2.36 m spatial resolution. The geometric accuracy of ZY-1 02C HRC is evaluated using two revisited data in Northwest China. Based on the ground control points (GCPs) extracted from the ZY-3 standard stereo triplet products, the adjustment accuracy is analyzed, the root mean square error (RMSE) of panchromatic images is within 2.0 pixels. Further, through static and dynamic residual analysis of GCP, it is found that due to the influence of the vibration of the satellite platform, there is a significant dynamic imaging error in the image data, especially in the push-broom direction along the orbit, an effect of about 5.0 pixels in size is detected from the image. After correcting the vibration, the geometric accuracy of the image can be improved to within 1.0 pixels, reaching the geometric accuracy of most images, reducing the adverse effects on the application. In general, this article quantitatively analyzes the geometric accuracy of the ZY-1 02C HRC image, studies the non-negligible vibration effects of the satellite platform, optimizes the geometric accuracy through the data processing method of attitude compensation, and improves the quality of image data.

Liping Zhao

River Shoreline Recognition from High Resolution Remote Sensing Imagery Based on Water Index and Artificial Neural Network

River shoreline recognition from high resolution remote sensing images is a hot spot in remote sensing and relevant research and application fields. Due to the complexity of river shorelines in high resolution remote sensing images, automatic recognition of the shorelines is still difficult, which needs further development. In this paper, a new intelligent method for river shoreline recognition was proposed, which was based on the integration of water index and artificial neural network. Key issues in the recognition algorithm were analyzed and the recognition algorithm was developed. Two representative areas of the Changjiang River and the Zhujiang River, were selected as the study areas. Corresponding GF-1 and GF-2 multispectral images were used as the experimental data. Visual comparison and quantitative assessments of the recognition results by the proposed method and the traditional methods were given. The experimental results show that the proposed method is effective and better than the traditional methods, which obtained most satisfactory recognition results of river shorelines.

Yuan Yao, Linyi Li

Collaborative Classification of Hyperspectral and LiDAR Data Based on Gram Matrices Constrained Fusion Net

Multi-sensor information collaborative utilization has attracted considerable attention in remote sensing area. While earth observation benefits from information diversity, multi-sensor collaborative classification technique is still confronted with varied challenges, including inconsistent data volume, different data structures, and uncorrelated physical properties. In this paper, a Gram matrices constrained fusion net (GMCF-Net) is designed for controlling multi-source heterogeneous information and improving classification performance. Exploiting the capability of the Gram matrix in capturing image textures, a multi-source structure control module is constructed to simultaneously address issues involved with different data volume and data structures, which matches the Gram matrices of multi-domains in a double-interweaving pattern. Finally, classification results are obtained based on the discriminative fusion from GMCF-Net. Extensive experiments built from two benchmark remote sensing data sets are reported, and the results demonstrate that the proposed framework yields state-of-the-art performance on hyperspectral and LiDAR data collaborative classification.

Mengmeng Zhang, Wei Li, Ran Tao

Joint Classification of Hyperspectral and LiDAR Data Using Improved Local Contain Profile

Joint classification using multi-source remote sensing data has drawn increasing attention. For some complex surveyed scenes, relying on a single hyperspectral image (HSI) is not enough to meet the purpose of high-precision classification. Comparatively, light detection and ranging (LiDAR) data is rich in structure and elevation characteristics. To extract discriminative features, state-of-the-art extinction profile (EP) and local contain profile (LCP) have been investigated. However, EP is more sensitive to the external environment such as shadow occlusion, and the filtering strategy in LCP is prone to lose useful information when dealing with complex terrain scenes. Therefore, an improved LCP (ILCP) method is proposed to extract features from HSI and LiDAR data for joint classification. The proposed ILCP is more stable than EP, and the dimension of features extracted by ILCP is half of EP, which can avoid Hughes phenomenon. Compared to LCP, ILCP uses threshold-based filtering instead of extinction value, which retains more useful information for classification. Furthermore, feature-level fusion is applied to extracted features, and then the integrated features are input to the support vector machine (SVM) for final classification. In this paper, EP and LCP are also employed as comparison methods. Experimental results validated with one real multi-resource remote sensing data demonstrate that the proposed ILCP is superior to traditional methods.

Dandan Cao, Wei Li, Lu Li, Qiong Ran, Mengmeng Zhang, Ran Tao

A High-Resolution Spotlight SAR Imaging Method Based on Two-Step Processing Approach

The modern radar system not only works on the conventional stripmap synthetic aperture radar (SAR) mode, but also on the high-resolution spotlight SAR mode with refined imaging. However, due to the azimuth spectral aliasing of echo signals in the spotlight SAR, the direct stripmap SAR imaging algorithm fails. To solve this problem, a high-resolution spotlight SAR imaging method based on deramp “two-step processing” is proposed. Firstly, the motion error is compensated based on the inertial navigation system (INS) and echo data, then azimuth deramp operation is performed to eliminates the azimuth spectral aliasing, and thus obtain the 2D signal spectrum without blurring. Then the modified Range-Doppler (RD) method is used to complete the range cell migration (RCM) correction and range compression. Finally, azimuth compression is performed by means of azimuth matched filtering to acquire the focused image. In this method, the imaging framework for both the stripmap SAR and the spotlight SAR are unified by the deramp operation without interpolation. The imaging results in the spotlight SAR mode have demonstrated the effectiveness of the method.

Hao Zheng, Hongmeng Chen

Research of Industrial Heat Sources Detection Method Based on Thermal Infrared Satellite Data

Carbon based fuel burning in industrial plants accounts for a large proportion of greenhouse gas emissions, which has a significant impact on human health and climate change. Monitoring the operation status of these factories by means of thermal infrared (TIR) remote sensing technology is one of the important measures for environmental protection. This paper proposed an automatic detection method for industrial heat sources (IHSs) based on TIR data. Firstly, the land surface temperature data is retrieved, then the local abnormal high temperature pixels are extracted by sliding window smoothing, and finally the industrial heat sources are further screened by normalized vegetation index (NDVI) data. To verify the reliability of the proposed method, this paper used the Landsat8 data in Jiangsu and Hebei Province of China in 2019 to identify industrial heat sources. Through artificial judgment with optical satellite image, the detection accuracy was 83.78% in Jiangsu and 78.6% in Hebei respectively. Whether for large-scale plants or small workshop plants, the detection accuracy of proposed method is satisfactory. The validation results represent that this method could be used as a new technology for government or industry regulation.

Jie Wang, Yu Liu, Hongzhong Lu, Jianwei Qi, Ping Ai

Robust Detection and Tracking Algorithm for Pedestrian Targets with Posture Change

In the intelligent monitoring system, pedestrian detection and tracking is the basis of behavioral analysis, pedestrian retrieval and other intelligent analysis technologies. In order to solve the problems caused by posture change and occlusion, this paper proposes a robust pedestrian target detection and tracking algorithm, which is based on YOLOv3 and DeepSORT. In order to reduce the impact of pedestrian posture change, we introduce a person re-identification feature that resists posture change on the tracking algorithm DeepSORT. For pedestrian occlusion, inspired by Spindle Net, we integrate seven local appearance features extracted from different regions of the human body together to form robust global appearance features, making the tracking more accurate. Experimental results prove that the algorithm proposed in this paper shows robust pedestrian detection and tracking performance on both subjective visual effects and objective metrics, which can meet the needs of actual applications.

Tao Zhao, Jin-Sheng Xiao, Wen Wang, Hai-Gang Sui, Jing-Jin Ma, Chuan Xu

Comparative Analysis of Chinese High-Resolution Satellite Data for Sugarcane Classification Based on U-Net Model

Monitoring sugarcane in China is important for the sugarcane industry which needs harvest progress information during the harvest season. The satellite image is one of the cost-effective and dynamic data sources for sugarcane classification recently. However, there are not many previous works for the classification of sugarcane with high-resolution satellite images especially sub-meter resolution data at present. Deep learning with a high performance of classification in agriculture was used in recent research. In this study, Chinese high-resolution satellite data based on the U-Net model was chosen to get a more precise segmentation of sugarcane in Laibin. GaoFen-1(GF-1) image with 2 m resolution and GaoFen-2(GF-2) image with 0.8 m resolution were compared. GF-2 image has a good performance in the OA and Kappa coefficient compared with the GF-1 image which shows that a high-resolution image can get better segmentation results of sugarcane than the low resolution using the same data and method. Furthermore, to get more precise results of sugarcane classification, two different growth stages of sugarcane GF-2 image were chosen: tillering period data in May and a grand growth period in August. The result shows the grand growth period is suitable for sugarcane classification with a better improvement in the OA and Kappa coefficient.

Chen Chen, Linjiang Lou, Xinyuan Gao, Yu Liu

A Differential Flat Approach for Attitude Manoeuver Control of Agile Satellite

This pater introduces a kind of control method for agile satellites which use SGCMG as actuators. Considering the SGCMG’s singularities, an attitude trajectory planning method is designed based on the differential flatness theory firstly. This method considers the singularity features and the influence of gravity gradient torque. A sliding mode tracking controller is proposed based on the Modified Rodrigues parameters. Result of simulation shows that the SGCMG undergoing large maneuver without singularity and the tracking error is less than 0.03°, the attitude deviation in the steady state is less than 1 × 10–5°.

Deting Li, Yikang He, Shuyu Lin, Miao Li, Kang Shi

A Voxel-Based Fusing Method for Aerial Laser Scanning and Oblique Image Point Cloud Via Noise-and-Occupancy-Aware

High resolution remote sensing is the main data source for urban 3D modeling, among which airborne laser scanning and oblique photogrammetry have become the two most important data acquisition methods. In the complex urban environment with tall buildings, these two methods have their own Pros and Cons. The airborne LiDAR point cloud with high precision is limited by the low spatial resolution and single top viewing angle, which fails to accurately describe the facade information of the building. The oblique image-based point cloud, which has the advantages of multiple views and rich texture, is limited by the noise, the sharp features cannot be accurately expressed either. The fusion of the two kinds of data brings new potential for breaking through the bottleneck of single data source. The completeness and accuracy of city description by point cloud data are expected to be improved, so as to provide accurate data support for target extraction, 3D reconstruction, spatial analysis and disaster protection. In view of the significant differences between the two kinds of point cloud data in spatial resolution, coverage and noise level, this paper proposes a cross-source point cloud fusion method that takes both mass data processing efficiency and local detail fidelity into account. First, the method is based on a robust registration algorithm, which extracts accurate point cloud subsets from image-based point clouds affected by noise, and corrects the initial registration results provided by inertial navigation system and image control points. Second, an adaptive voxel segmentation method based on plane structure perception is proposed to establish correspondence of cross-source point clouds. Finally, cross-source point clouds are filtered for mutual exclusion. Experiments show that the proposed method integrates the two groups of point clouds, and the obtained results complete the vacancy of the LiDAR point cloud and improve the accuracy of the image-based point cloud.

Shiming Li, Qing Zhu, Han Hu, Xuming Ge, Chuncheng Zhu

A Fast and Robust UAV Images Mosaic Method

Unmanned Aerial Vehicle (UAV) can respond quickly and provide image data at the first time in mapping support tasks, such as battlefield environment monitoring and natural disaster damage assessment. However, UAV is prone to fast flight speed and poor attitude stability, which brings a new challenge to image mosaic relying on feature matching. In this paper, an efficient and robust UAV image mosaic method is proposed. The main idea is to take graph theory and POS data to image mosaic process. Firstly, we estimate the range of aerial photography based on POS data, and construct the UAV images relationship diagram. Then, according to the attitude information, the images in relation set are divided into stable group and disturbance group, and the image classification connection graph is generated. By defining the weighted topological graph and constructing the minimum spanning tree, we search key frames and reference image. Finally, we derive the homography matrix between adjacent images according to imaging model. The mosaic image is generated after global registration and images fusion. Experimental results show that our proposed approach can solve the mosaic problems of narrow overlap and texture deficient images, and produce superior results in reducing projection distortion and shortening registration time consumption.

Guangrui Yu, Changliang Ha, Chunlin Shi, Lianbing Gong, Lili Yu

Comparative Research on Water Body Extraction Methods Based on SPOT Data

With the rapid development of science and technology, it has become an inevitable trend to use remote sensing technology to monitor, investigate and analyze urban waters. The rapid and accurate extraction of water bodies from remote sensing images has become an important water extraction method. Based on SPOT4 images, this paper applied Ostu Threshold method and Maximum Likelihood Classification method to conduct experiments on water body extraction and compared the advantages and disadvantages of supervised classification and unsupervised classification in urban waters area extraction. Experimental results show that the Maximum Likelihood Classification method has the highest accuracy, and the Threshold Method has the fastest efficiency, but small rivers cannot be distinguished.

Linjiang Lou, Chen Chen, Xinyuan Gao, Kun Liu, Minmin Li, Yajie Fu

Research on National Land Resource Accounting Method

The accounting on national natural resources is the basis of acknowledgement of the asset value of our national resources, which is also the important criterion on the supervision, examination and management of national resource assets owned by the whole people of China. Land resource is not only the significant component of the natural resources, but also the spatial carriers of the natural resources such as mineral, forest, grass, wet land and water. In recent thirty years, the implementation of the system of paid use of land has made the value of the assets of the land resources more and more explicitly, which has laid more and more economic and social significance on the state governance, macro economy, national wealth, and the property owners. The content, method of accounting on land resources is the key point of the accounting on natural resources. This essay brings up the accounting method based on the physical quantity and value of the land resource assets. The physical quantity accounting could be relied on the basic data of the accounting area, and be calculated based on the forward operator, backward operator, and forward and backward operator method. The value accounting, based on the benchmark land price of the accounting area, could be measured on the comprehensive value on the whole accounting area weighed by the area at different levels. Though this accounting method is practicable, the problems in the management process need to be taken seriously. The conclusion is that the quality of the national land resource asset accounting relies on the solid data fundamentals and the relatively matching managing system.

Xinyan Zheng, Yu Zhou, Tao Cheng, Jin Liu, Chen Chen, Lu Yi

A Configuration Approach and Effectiveness Evaluation of Modular Reconfigurable Satellite

Modular reconfigurable satellites are characterized by low cost and short cycle of development, launch, operation and maintenance, which can maximize the interests of developers and operators while meeting the needs of users. However, this field is still in its infancy and has huge market potential. On the basis of the domestic and international research on modular reconfigurable satellite structure and technology, this paper proposes a new configuration of modular reconfigurable satellite. The key technologies, operation and maintenance modes, and commercial application prospects are studied according to the configuration. In this paper, a modeling of this modular satellite configuration is established in a simulation system, and its performance in earth observation application scenarios is evaluated by means of simulation data and evaluation index system. The evaluation results show that this configuration can improve the effectiveness of modular reconfigurable satellites.

Fu Xin, Mai Xiamei, Zhang Haolong, Xue Fengtong

FD-LinkNet: A Encoder-decoder Structure Network for High Resolution Satellite Imagery Rural Road Extraction

Due to the backward construction of rural road infrastructure and lack of supervision, unreasonable use of land and frequent safety accidents have occurred. At present, road extraction is a hot task in the area of high resolution satellite imagery target sensing. It is the basis of rural road management to extract rural road through the high resolution satellite imagery. Based on this, the state of art road extraction semantic segmentation neural network named D-LinkNet is applied and improved in this paper. The improved semantic segmentation neural network is named FD-LinkNet. The FD-LinkNet network uses encoder-decoder architecture and adds dilated convolution layers in center part. The encoder part takes advantages of ResNet34 pretrained on ImageNet dataset. Dilated convolution layers with skip connections is applied to full scale to enlarge the receptive filed of feature points without reducing the feature maps. Finally, the decoder part output the binary images as the binary semantic segmentation results. Compared with UNet, and D-LinkNet, The best IoU scores on the validation set and test set of FD-LinkNet are 0.6819, which has increased by 8.3%, and 5.5% respectively. The trained network in this paper can effectively extract rural road, which is benefit for road management departments to detect and superbise rural road.

Lin Yi, Miao Yang, Liang Shuang, Peng Xiangyang, Song Wentao

Geolocation Accuracy of Bistatic InSAR Configuration with Geostationary Transmitter and LEO Receivers

Spaceborne GEO-LEO bistatic InSAR system with broad application prospects is an effective approach to obtain high-precision digital elevation model (DEM). Nevertheless, the geometry of bistatic interferometry is complex compared with traditional InSAR system, meanwhile the range equation is in the form of round-trip delay. In this paper, the InSAR geolocation equations, based on the GEO-LEO bistatic geometry, are established by combining SAR Doppler equation, phase equation and range equation. Furthermore, the geolocation errors caused by all error sources are derived by calculating the partial derivatives. Simulation experiments are carried out to analyze the geolocation errors, which can obtain meter-level precision and mainly caused by the absolute interferometric phase error, the range error and the baseline measurement error.

Yuekun Wang, Zhonghao Wei, Feng Tian, Long Zhuang

Preliminary Evaluation of the Stereo Mapping Accuracy of the Gaofen-7

The Gaofen-7 (GF-7) is China’s new-generation sub-meter stereo mapping satellite and an important part of the CHEOS (China High Resolution Earth Observation System). The stereo mapping camera carried on GF-7 allows acquisition of along-track stereoscopic pairs with a swath width greater than 20 km, a resolution better than 0.8 m, and a base-to-height ratio of approximately 0.6. GF-7 stereo data over Sichuan and Heilongjiang test sites are used for the evaluation and analysis of 1:10,000 scale stereo mapping accuracy. The test results show that, according to China’s mapping specifications, products such as digital line graphic (DLG), digital terrain model (DTM), digital surface model (DSM) and digital orthophoto map (DOM) can meet the accuracy requirements for hilly, mountainous and alpine terrain. For the flat terrain type with the highest accuracy indicators, products such as DTM, DSM and DOM can also achieve the required accuracy; but for the DLG products, the elevation accuracy is not always able to meet the specifications, which need to be further studied.

Liping Zhao, Zhi Fang, Qifeng Chu

Research on Cloud Detection Method of GaoFen-6 Wide Camera Data

Cloud detection is one of the research hotpots in high-resolution remote sensing images, and is an important part of remote sensing image preprocessing. The traditional physical threshold method based on pixel is inefficient when the amount of data is large. It is also easily affected by the noise points in the image, and the appropriate spectral threshold is difficult to determine. In view of the above problems, this paper starts from the GF-6 wide camera data, studies the spectral characteristics of GF-6 wide camera data, and selects the appropriate spectral detection band. Then, in order to improve the universality, using the maximum inter class variance method (OTSU) to adaptively obtain the appropriate cloud detection threshold. Finally, the image is segmented by simple linear iterative clustering algorithm (SLIC). Super pixel is used as the processing unit of subsequent image processing to replace single pixel for cloud detection. It both improves validity and accuracy. In this paper, by doing experiments and quantitative evaluation, it can be proved that the cloud detection algorithm can effectively and accurately detect the cloud region in the image. The accuracy was all higher than 90%, and the average accuracy of the experiment is 94.21%.

Shiyun Ke, Mi Wang, Jinshan Cao, En Long

Research on Space-Ground Integrated Control Technology of Earth Observation Spacecraft

With the improvement of spacecraft processing capacity, spacecraft has gradually possessed certain autonomous management capability. At present, users have put forward higher and higher expectations on the ease of use of earth observation spacecraft. One task often needs to schedule multi earth observation spacecraft of different types, scales and spectra. But the multi spacecraft task scheduling mainly relies on the cooperative mode of “spacecraft ground spacecraft”, which has the disadvantages of poor timeliness, complex user interface and low cooperation efficiency, which cannot meet the needs of users. To solve this problem, this paper proposes an integrated autonomous satellite group management and control technology, which can coordinate the relationship between ground control and spacecraft autonomous management and control capability through hierarchical task planning, space resource synchronization and joint health assessment, so as to improve the joint control effect and realize the rapid response of spacecraft payload business, which can be used as a reference for future spacecraft mission management and control design.

Zhang Fajia, Yang Tongzhi, Zhu Jia, Xing Nan, Dang Jiancheng

An Improved Ionospheric Estimation Algorithm for Low-Frequency SAR Interferometry

Interferometric Synthetic Aperture Radar (InSAR) has been proven to successfully map surface deformations and monitor glacier movement. For space-borne InSAR systems working at low frequencies, the deformation measurement accuracy is strongly affected by the ionospheric effect, which can cause phase delay and lead to an obvious interference phase error. For compensation of the ionospheric effect, an improved ionospheric estimation algorithm based on the split spectrum is proposed. The theoretical basis and implementation process of the improved algorithm are described in detail in the article. In order to investigate the estimation accuracy of the proposed algorithm, simulation data with complex ionospheric changes are used and six different filtering algorithms are selected for optimization of the proposed algorithm. According to the results of the ionospheric estimation error, the improved algorithm has a smaller mean value and standard deviation than the traditional algorithm. Among the six filtering methods, the mean value and standard deviation of the ionospheric estimation result error of the Lee filtering method are much smaller than those of the original method. These results illustrate that the proposed algorithm improves the actual accuracy of ionospheric estimation and strongly certifies the effectiveness of the improved algorithm.

Beixin Qin, Yongsheng Zhang, Zaoyu Sun, Xiaoxiang Zhu

The Research of Multi-scale Effect on Remote Sensing Image Object Detection

Object detectors based on convolutional neural network (CNN) have already achieved state-of-the-art performance due to the powerful feature representation capabilities in computer vision tasks including image classification, object detection, and image segmentation. With the rapid development of remote sensing technology, the object detection of high-resolution optical remote sensing images has become an important part of remote sensing. On account of the large sight and complex background of the remote sensing images, along with the large size variation of the objects, the performance of the methods based on CNN is restricted. This paper will focus on the objects that multi-scale features are of the most obvious to analyze and discuss the impact of multi-scale effect on object detection. A new criterion for setting small objects is presented. Through the analysis of evaluation indicators, we conclude that the multi-scale effect decreases the detection accuracy of small objects under the same condition. Furthermore, a method is proposed to improve the accuracy of small objects detection.

Guangkuo Ma, Yifan Dong, Yun Su, Wei Xu, Pingping Huang

Remote Sensing Road Segmentation Based on Feature Extraction Optimization and Skeleton Detection Optimization

High-resolution remote sensing image segmentation is a mature application in many industrial-level image applications, such as those in the military, civil, and other fields. Scene analysis needs to be automated in high-resolution remote sensing images as much as possible. Nowadays, with the rise of deep learning algorithms, remote sensing image processing algorithms have made tremendous progress. Deep learning algorithms process unlabeled data by learning a certain amount of labeled data. We conducted a specific study on the road target with GF1 data collected in China, and the remote sensing image’s resolution was 2 m (Jin et al. in J Anhui Agri Sci 43:358–362, 2015 [Jin et al. in J Anhui Agri Sci 43:358–362, 2015]). According to the observation of road features in remote sensing images, it still has a large number of small roads that are difficult to distinguish in the 2 m resolution GF1 remote sensing image. Due to the limitations of the downsampling calculation of the fully convolutional neural network, it is easy to lose a lot of information on small roads (Long et al. in Proceedings of the IEEE conference on computer vision and pattern recognition, pp. 3431–3440, 2015 [Long J, Shelhamer E, Darrell T (2015) Fully convolutional networks for semantic segmentation. In: Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp 3431–3440]). Therefore, we have adjusted the feature extraction and backbone networ. We adopted EfficientNet (Tan and Le in Efficientnet: Rethinking model scaling for convolutional neural networks, 2019 [Tan M, Le QV (2019) Efficientnet: rethinking model scaling for convolutional neural networks. arXiv preprint arXiv:1905.11946]) as the skeleton network of the algorithm, and combined D-linkNet (Zhou et al. in CVPR workshops, pp. 182–186, 2018 [Zhou L, Zhang C, Wu M (2018) D-LinkNet: LinkNet with pretrained encoder and dilated convolution for high resolution satellite imagery road extraction. In: CVPR workshops, pp 182–186]) as our After preliminary training to proposed D2-LinkNet, based on the road samples, we can get very good prediction results. According to the existing prediction results, there is still a certain difference in the fitting of our segmentation results to the groundtruth. To solve this problem, based on the extraction and analysis of the skeleton of the road prediction results and the integration of different prediction results, we proposed a skeleton optimization algorithm to optimize our prediction results for road samples. And on the one hand, it complements the segmentation of small roads. On the other hand, the road segmentation result fits the road boundary more closely. The final experiment results show that the optimization algorithm effectively improves the prediction accuracy of the network, and we have achieved better results compared with D-LinkNet in our large range remote sensing dataset.

Sijun Dong, Yu Zhao, Zhengchao Chen

Capability Requirement Analysis of Space-Based Ocean Reconnaissance and Surveillance System Based on Improved QFD Method

The capability requirement analysis of space-based ocean reconnaissance and surveillance system is the basic work of system design, which has an important influence on the construction and optimization of system capability. By summarizing the concept and application status of QFD (Quality Function Deployment) method, and combining with the characteristics of space-based ocean reconnaissance and surveillance system capability requirement analysis, an improved QFD method based on fuzzy theory and GRA (Grey Relational Analysis) is proposed. Based on the analysis of space-based ocean reconnaissance and surveillance task and capability composition, the HoQ (House of Quality) model is established by using the improved QFD method. Through case analysis, the importance ranking of each capability is obtained. The research results verify the feasibility of the improved QFD method, and can also provide reference for the design and optimization of space-based ocean reconnaissance and surveillance system.

Min Wan, Yan Hou

Remote Sensing Scene Classification with Multi-task Learning

Remote sensing scene classification is a fundamental task for understanding the high-resolution remote sensing images and has achieved great progress thanks to the deep neural networks. However, the deep neural networks heavily depend on the large-scale training samples, while sufficient manually annotated data is often unavailable. In this paper, a Multi-Task Scene Classification Network (MTSCN) is proposed to address the label consumption dilemma by using annotated information from other related tasks. The MTSCN simultaneously conducts multiple tasks so that scene classification task can leverage information shared by other tasks. The MTSCN consists of a shared branch and multiple task-specific branches with each for one task. First, the shared branch extracts shared features for all tasks. Then, the task-specific branch adapts shared features into task-specific features to finish the classification task. Such architecture allows complementary information shared among tasks as well as avoids task interference by selecting specific information for a different task. The method is evaluated on two multi-task learning scenarios with respect to accuracy and the number of network parameters. The results show that the proposed method is efficient not only in terms of data but also in terms of network parameters.

Tengfei Gong, Xiangtao Zheng, Xiaoqiang Lu

Top-of-Atmosphere Radiance Image Simulation from the Visible to Thermal Infrared

Top-of-Atmosphere (TOA) radiance image simulation is very useful for the definition of new observation systems, the optimization of optical instruments and the design of data processing or scientific algorithms. A physical based simulation model from visible to thermal infrared incorporating the coupling between heterogeneous non-Lambertian surface and atmosphere is proposed in this paper. It is necessary for multi-angle optical remote sensing of natural land. The simulation examples in optical domain and thermal infrared domain are analyzed. At last, we suggest the further investigation about the acquisition of high spatial and high spectral resolution surface Bidirectional Reflectance Distribution Function (BRDF), which helps the proposed model applied to various scene generation demands.

Dong Liu, Yanbing Dong, Hongxia Mao

Application of Improved RBF Neural Network in Remote Sensing Image Restoration

Neural network is widely used in remote sensing image restoration. To avoid huge time and space consumption of conventional neural network models, a novel remote sensing image restoration algorithm base on improved RBF (Radial Basis Function) neural network is proposed. The training data set could be organized by degrading the high-quality remote sensing images or promoting degraded images’ quality by other necessary methods. And the number of hidden layer is decided by the size of training data at the condition of small training sample scale. To accelerate convergence speed of RBF neural network during training process, conjugate gradient descent method is adopted to realize the weight parameters’ iterative correction. To further reduce calculating time, this paper proposes a matrix factorization algorithm to realize matrix's parallel arithmetic. Simulations and experiments indicate that the improved RBF neural network model could acquire relatively approving remote sensing image restoration results and time overhead.

Yunsen Wang, Yong Wang, Daqiang Feng, Jing Yang, Ke An

Distributed HRWS MIMO-SAR Linear Vertical Baseline Estimation and Compensation

The satellite formation configuration directly determines the functions of the distributed SAR system and has an important impact on system performance. The most practical formation configuration for realizing high resolution wide swath imaging is satellite formation along track. However, during the formation flight, the distributed array changes with time, so there is a linearly changing vertical baseline. The linearly changing vertical baseline will cause mismatch of the steering vector and cause the azimuth ambiguity resolution to fail. Based on the satellite formation configuration along track, this paper studies the vertical baseline variation and its influence on the ambiguity resolution along the track and converts the vertical baseline into a phase term for estimation and compensation. Through experimental analysis, by estimating and compensating the vertical baseline, the ambiguity caused by the mismatch of the steering vector caused by the linearly varying vertical baseline can be effectively solved.

Pengcheng Li, Zaoyu Sun, Yongsheng Zhang, Feng He, Shenjing Wang, Zhen Dong

Fast Three-Dimensional Imaging Method of Aerospace Targets Based on Linear Array Radar

Distributed Array Radar (DAR) uses space diversity to obtain better detection performance. As a new configuration of DAR, Linear Array Radar (LAR) can achieve a higher frame rate and higher dimensional imaging of aerospace targets. Based on the real-virtual aperture hybrid mode, LAR obtain the three-dimensional image of the target through space–time dimensions. The imaging process using the three-dimensional Back Projection (BP) algorithm accurately reconstruct the three-dimensional image of the target, but the amount of calculation is huge. In order to solve this problem, the Fast Factorized Back Projection (FFBP) algorithm is extended to three-dimensional space. And in the Cartesian coordinate system, a detailed theoretical derivation of the subaperture coarse image interpolation method is made. Through the imaging experiment and imaging quality evaluation of ideal point targets, it is verified that the fast imaging algorithm has good imaging quality. It has only a small performance loss compared with the traditional three-dimensional BP algorithm. The imaging processing time is also greatly reduced. In addition, this paper conducts imaging experiments based on satellite model simulation data generated by Radarbase. The results show that the algorithm can reconstruct the geometric configuration of the target well, which verifies the effectiveness of this algorithm.

Zhang Yi, Yu Anxi, Jin Guanghu, Dong Zhen

Edge Detection of SAR Images Based on Shearlet

Synthetic aperture radar (SAR) has the characteristics of all-weather imaging and is widely used in various fields. However, owe to coherent imaging mechanism, speckle appears in SAR images, which seriously affects the interpretation of images. Due to merits of the strong directional sensitivity and the optimal sparsity, shearlet is used to construct an edge detector for SAR images. The detector firstly converts the constructed even symmetric shearlet into odd symmetry by Hilbert transform, then determines the main direction of the edge according to the odd symmetric shearlet coefficient of the SAR image, and calculates the possibility of edge existence according to both odd and even symmetric shearlet coefficient. Experiments show that, in comparison with the traditional difference-based edge detectors, the proposed edge detector has stronger ability of anti-speckle interference, and can better detect the edge information in SAR images.

Zengguo Sun, Guodong Zhao, Weirong Chen, Robertas Damaševičius, Marcin Woźniak

Research on Hysteresis Control Method of Fast Steering Mirror in Satellite Laser Communication

In the precise pointing system of satellite laser communication, the piezoelectric fast steering mirror (FSM) is used in the system. In the free state, the hysteresis between the input voltage and the output displacement of FSM is obvious, which has a serious impact on the precision positioning accuracy. Therefore, aiming at the hysteresis problem of the fast steering mirror in satellite laser communication, a phenomenal hysteresis mathematical model based on Bouc-Wen operator is established by analyzing the hysteresis mechanism of the piezoelectric fast steering mirror. An improved nonlinear particle swarm optimization (PSO) parameter identification method is proposed, which greatly improves the optimization efficiency of particle swarm optimization. On this basis, a feedforward inverse compensation linearization method is proposed. In order to verify the effectiveness of the proposed model, a fast steering mirror system experiment is established. The experimental results show that the improved PSO parameter identification method proposed in this paper can accurately identify the parameters in the model, and the proposed feedforward linearization method can improve the linearity of piezoelectric ceramic actuator by more than 90%, which can meet the actual index requirements of the fast steering mirror.

Yu Zhiliang, Li Xue, Yang Mingliang, Fang Zhiyi

Construction of Remote Sensing User Growth System Based on Hooked Mode

Remote sensing information has been widely used in various industries of national economy. In order to promote the benign ecological development of remote sensing information producers and users, it is necessary to build a growth system of remote sensing users. This paper analyzes the common methods of typical Internet user growth system, and summarizes the essence of user growth system. Aiming at the construction of remote sensing user growth system, firstly, the target users, application scenarios, product concepts, service platforms and strategic directions of remote sensing information are defined. Then, the driving force based on hooked model is designed, and the RFA model suitable for the growth of remote sensing users is constructed. Finally, a closed-loop remote sensing user growth system is constructed by taking the integral method as an example, so that users can continuously apply remote sensing information, Promote the continuous development of remote sensing means and form a virtuous circle.

Caiping Li, Xiaoming Zhou, Jisheng Zhang, Lei Chang

Ship Azimuth Velocity Estimation in Spaceborne SAR Based on Minimum-Entropy Criterion and Newton’s Method

Spaceborne synthetic aperture radar is of great significance in ship monitoring. Estimation of motion parameters of moving ships has a high application value in civil and military fields. In this paper, we propose a method to estimate azimuth velocity of moving ships in spaceborne SAR as well as improve the image quality. This method is applied directly on the single-look complex (SLC) image data, and does not demand extra information or special configurations. Applying Newton’s method to search the calibration phase corresponding to the minimum of the entropy, the Doppler frequency modulation (FM) rate and azimuth velocity of ships can be obtained by fitting the quadratic coefficient of the correction phase, and the image blur caused by the movement of ships is effectively alleviated by means of phase compensation. Finally, the proposed method is validated by the Gaofen-3 (GF-3) complex data, and the experimental results perfectly fit the data recorded by the international automatic identification system, which explains the effectiveness of the proposed approach.

Yating Huang, Dexin Li, Zhen Dong

Mapping 30-m Resolution Land Cover of China Based on Full Convolutional Neural Network

Large-scale land cover maps are essential geoscience information for many applications. However, the cost of the current mapping process greatly limits the update frequency of large-scale land cover products. Moreover, the inconsistency among land cover maps also hindered further long time-series land cover change analysis and research. Therefore, we proposed a novel framework to efficiently generate new land cover maps using historical land cover products. The framework mainly includes the following three innovations: (1) To solve the influence of clouds and seasonal inconsistency, use Google Earth Engine to composite the multi-temporal remote sensing images to generate consistent cloud-free images; (2) To avoid the repeated collection of training samples, training data were generated by integrating multiple historical products; (3) To process massive generated training data and improve mapping accuracy, the data-driven deep fully convolutional network model is used to achieve end-to-end land cover mapping. Based on the proposed mapping approach, the 30-m resolution land cover map of China in 2015 was automatically completed with improved accuracy, which shows the potential for frequent large-scale land cover product integration and updating.

Yinhe Liu, Yanfei Zhong

Research on Brightness Compensation Algorithm for GF-3 SAR Images

GF-3, the first C-Band full-polarimetric synthetic aperture radar (SAR) satellite with a space resolution up to 1 m, has multiple strip and scan imaging modes. In this paper, a brightness compensation model is proposed to correct the unbalanced radiation image of GF-3 satellite in some areas based on mathematical morphology principle, SAR imaging model, and histogram mapping model, so that the image radiation is relatively consistent and has moderate contrast. The proposed algorithm is tested on the GF-3 SAR image with radiation difference in Taklimakan desert area of Xinjiang, and is compared with the MASK dodging algorithm in terms of brightness mean, standard deviation, information entropy, and average gradient. Results of brightness compensation show that, the proposed algorithm makes the radiation differences within images smaller, and the image quality is obviously improved, which provides a non-radiation difference image for subsequent applications.

Zengguo Sun, Xiaopeng Yan, Qianfu Chen, Weirong Chen

The Spaceborne SAR 3D Imaging Method Based on 2D Compressed Sensing

Synthetic Aperture Radar (SAR) three-dimensional (3D) imaging technology uses multiple flight observations in the cross-track direction, to obtain three-dimensional spatial distribution of observation scenes. In order to shorten the revisit time of the satellite and increase the number of orbits, this paper proposes a sparse 3D imaging method of two-dimensional (2D) Compressed Sensing (CS). Under the large orbit distribution and sparse flight, the linear measurement matrix in range and cross-track directions is established, which is conducive to the joint solution of CS theory for images under sparse representation and to avoid the coupling of echo signals in the orbit and distance directions. The simulation results of imaging resolution of 0.5 m for single satellite and double satellites are given, verifying the effectiveness of the method.

He Tian, Chunzhu Dong, Hongcheng Yin
Weitere Informationen