Table of Content

20 September 2023, Volume 52 Issue 9

Previous Issue    Next Issue

Review

Research progress of geodesy in China (2019—2023)

DANG Yamin, JIANG Tao, YANG Yuanxi, SUN Heping, JIANG Weiping, ZHU Jianjun, XUE Shuqiang, ZHANG Xiaohong, YU Baoguo, LUO Zhicai, LI Xingxing, XIAO Yun, ZHANG Chuanyin, ZHANG Baocheng, LI Zishen, FENG Wei, REN Xia, WANG Hu

2023, 52(9):  1419-1436.  doi:10.11947/j.AGCS.2023.20230343

Asbtract ( )   HTML ( )   PDF (1294KB) ( )  

References | Related Articles | Metrics

From July 11 to 20, 2023, the 28th International Union of Geodesy and Geophysics (IUGG) general assembly was held in Berlin, Germany. According to the tradition of IUGG, the Chinese National Committee for International Association of Geodesy (CNC-IAG) organized dozens of domestic institutions to compile the “2019—2023 China National Report on Geodesy”, which summarized the research progress of various branches of geodesy in China from 2019 to 2023. This article summarizes the overall progress of China's geodetic discipline in recent years, focusing on representative progress in six research directions including reference frame, comprehensive PNT and resilient PNT, gravity field and vertical datum, precise GNSS products, multi-source sensor integrated navigation, and marine geodesy. Moreover, in light of the development trends of international geodesy and related interdisciplinary disciplines, several suggestions are proposed for the future development of geodesy in China.

Geodesy and Navigation

Precise orbit determination using satellite laser ranging and inter-satellite link observations for BDS-3 satellites

QU Weijing, HUANG Yong, XU Junyi, SUN Shuxian, ZHOU Shanshi, YANG Yufei, HE Bing, HU Xiaogong

2023, 52(9):  1437-1448.  doi:10.11947/j.AGCS.2023.20220558

Asbtract ( )   HTML ( )   PDF (5587KB) ( )  

References | Related Articles | Metrics

BDS satellites are equipped with the inter-satellite link (ISL) equipment, which can observe other satellites and ground monitoring stations with Ka-band measurements. The relative satellite clock and geometric distance can be separated to decouple the satellite orbit and clock difference using the dual one-way ISL ranging measurements of BDS satellites. The geometric distance is taken as the observation and is combined with the ground-based measurements to determine the precision orbit of BDS satellites. Satellite laser ranging has no carrier phase ambiguity and clock difference and is not affected by the ionosphere. SLR data and data processing are relatively simple, which can be used as a measurement technology independent of GNSS. This paper focuses on precision orbit determination for 11 BDS satellites (MEO/IGSO/GEO) based on SLR and ISL measurements. The accuracy of the BDS-3 satellite orbit determination based on the SLR and the inter-satellite links is greatly improved compared with the SLR data only, especially for GEO and IGSO satellites. The satellite orbit accuracy of the three orbit types is equivalent. The accuracy is 4.2 cm for radial component and 30.2 cm for 3D position. The accuracy of 12 h and 24 h predicted orbit is about 40.0 cm for the 3D position for MEO satellites, less than 60.0 cm for IGSO satellites and about 1 m for GEO satellites. The Earth rotation parameters are estimated simultaneously with the satellite orbit. Although the accuracy of the pole motion and LOD is about 3.0 mas and 0.35 ms due to the small amount of SLR observation data, the method is feasible to calculate the earth rotation parameters. The results show that the high precision orbit of the navigation satellites can be obtained using ISL measurements and small amount of the SLR data. If BDS constellation can be intensively tracked, it will help to improve the orbital accuracy. The results can also provide a reference for the solutions of BDS spatial datum parameters.

Undifferenced and uncombined PPP-RTK aided by BDS-3 PPP-B2b precise orbits

ZHA Jiuping, ZHANG Baocheng, LIU Teng, ZHANG Xiao, HOU Pengyu, YUAN Yunbin, LI Zishen

2023, 52(9):  1449-1459.  doi:10.11947/j.AGCS.2023.20220259

Asbtract ( )   HTML ( )   PDF (7877KB) ( )  

References | Related Articles | Metrics

Standard precise point positioning (PPP) service has been provided to users in Asia-Pacific region for free by the BDS-3 through B2b signals of BDS-3's geostationary earth orbit satellites (PPP-B2b), but its long convergence time of approximately 30 minutes and decimeter-level positioning accuracy in real-time processing are not conducive to its subsequent application and promotion. Therefore, this study proposes an enhanced positioning method that integrates precise satellite orbit products of PPP-B2b and observations from the regional sparse reference network, namely the undifferenced and uncombined PPP real-time kinematic (UDUC PPP-RTK) based on the PPP-B2b, and it is further constrained by the between-station single-difference ionospheric delay pseudo-observations to achieve a tight estimation of parameters such as ionospheric delay parameters. Besides, the single-satellite real-time modeling schemes of regional ionospheric slant delays and their accuracy information are emphasis designed in this study, which effectively compresses the amount of broadcast data while improves the application performance of PPP-RTK. Then, some near-real-time experiments have been carried out based on the Beijing-Tianjin sparse reference network to verify above theoretical method. The results show that the correction accuracy of the ionospheric slant delays calculated by the proposed method are 2.2 cm for BDS-3 and 2.4 cm for GPS; the absolute positioning errors of BDS-3+GPS converge to 2 cm (horizontal) and 5 cm (vertical) within 2 s for above 95% of arcs; and the millimeter- (horizontal) and centimeter-level (vertical) positioning accuracy can be achieved after convergence of positioning errors.

BDS-3 satellite difference code bias estimation with satellite phase center offset correction applied

LI Yang, WANG Ningbo, LI Zishen, WANG Liang, LI Zongyi

2023, 52(9):  1460-1468.  doi:10.11947/j.AGCS.2023.20220444

Asbtract ( )   HTML ( )   PDF (6916KB) ( )  

References | Related Articles | Metrics

The precise different code bias (DCB) correction information is basically required in the high-precision applications of multi-frequency GNSS. It is noted that the phase center offset (PCO) errors have not yet been properly handled in the generation of GNSS DCBs. In this paper, we first checked the variation characteristics of satellite PCOs of BeiDou global navigation satellite system (BDS-3), and analyzed the PCO effects on the generated BDS-3 DCBs. The empirical PCO correction model for DCBs (i.e., PCO-corrected-DCB)is then proposed, and the DCB estimation method with PCO correction applied (i.e., PCO-estimated-DCB) is also presented. Using BDS-3 observation data from the International GNSS Service (IGS) stations, the BDS-3 C2I-C6I/C1P-C5P DCBs with/without PCO corrections are estimated. The BDS-3 C2I/C1P single-frequency standard point positioning (SF-SPP) utilizing precise satellite orbits and clocks is performed to check the quality of the generated DCBs. Results show that the differences between DCBs estimated with and without PCO corrections reach 0.60 ns. The DCB discrepancy between different satellite types of BDS-3 is up to 1.17 ns, indicating the PCO errors in the generated DCBs cannot be ignored in the associated positioning applications. Compared to the BDS-3 SF-SPP result applying DCBs without PCO corrections,the accuracy improvement of SF-SPP based on PCO-estimated-DCB and PCO-corrected-DCB is comparable. The positioning accuracy improves by 5.7% and 6.8% in horizontal and vertical components, respectively, for PCO-estimated-DCB and PCO-corrected-DCB generated solutions.

A LiDAR/IMU spatial calibration method based on LiDAR labels and occupancy grid map

QIAN Chuang, ZHANG Hongjuan, LI Wenzhuo, LIU Hui, LI Bijun

2023, 52(9):  1469-1479.  doi:10.11947/j.AGCS.2023.20220242

Asbtract ( )   HTML ( )   PDF (25637KB) ( )  

References | Related Articles | Metrics

LiDAR and inertial measurement unit (IMU) have been widely used in intelligent vehicles, such as high-precision map generation, real-time vehicle positioning, etc. When LiDAR and IMU work together, it is necessary to know the spatial relationship between the two sensors, including the spatial rotation and translation parameters. This paper proposes an automated LiDAR/IMU spatial calibration method based on LiDAR labels. We first analyze the influence of LiDAR/IMU calibration parameters on LiDAR point cloud splicing, and prove that when the vehicle moves approximately in a straight line, LiDAR points are converted to a consistent axial direction using the approximate calibration parameters of LiDAR/IMU. Based on this conclusion, a method for generating a LiDAR grid occupancy map with high relative accuracy based on IMU attitude constraint is proposed. Point clouds of LiDAR labels with high-precision global position are matched with the map to obtain positions of the labels in the map. Based on the known high-precision position of the LiDAR labels, a nonlinear optimization method is used to solve the spatial transformation relationship between the grid occupancy map and the LiDAR labels, and the spatial calibration parameters of the LiDAR/IMU is further solved. The experimental results show that the point cloud map constructed by the solved LiDAR/IMU calibration parameters can achieve absolute position accuracy of centimeter level, which verifies the feasibility of our method.

Regularization parameter determination method based on MSE relative variation rule and its application in PolInSAR surveying inversion

LIN Dongfang, YAO Yibin, ZHENG Dunyong, LIAO Mengguang, XIE Jian

2023, 52(9):  1480-1491.  doi:10.11947/j.AGCS.2023.20220453

Asbtract ( )   HTML ( )   PDF (19393KB) ( )  

References | Related Articles | Metrics

The regularization method is currently the most widely used method for solving ill-posed problems in geodesy, and the regularization parameter is the key parameter that affects the solution result of the regularization method. With sufficient theoretical basis, the regularization parameter determination method based on the minimum mean square error (MSE) criterion can increase the estimation accuracy of model parameters efficiently. However, the calculation of the mean square error requires the true value of model parameters which is replaced by the estimated value in practice. As a result, the accurate mean square error is difficult to obtain, which greatly limits the effectiveness of the regularization parameters. In view of this, this paper analyzes the variation law of variance and bias caused by the changes of regularization parameter, and proposes a determination method for relative variation of mean square error. According to the principle that the true value of model parameters does not change under different regularization parameters, the calculation of the relative changes of variance and bias under different regularization parameters can effectively remove the influence of unknown true values of model parameters on mean square error estimation. This paper firstly uses different regularization parameters to calculate the relative changes of variance and standard deviation between the two regularization parameters; then calculates the model parameter estimate change between the two regularization parameters. The relative variation of bias under the two regularization parameters is obtained by difference operation analysis of variance change and model parameter estimate change. Finally, the regularization parameter with the maximum reduction of the mean square error is obtained by integrating the changes of standard deviation and bias. The feasibility of the new method is verified by the polarimetric interferometric synthetic aperture radar (PolInSAR) vegetation height inversion experiment. All Experiments show that the new method can effectively enhance the parameter estimation accuracy of the regularization method. Both of the parameter inversion accuracy of the two PolInSAR surveying experiments are improved. Those reasonably verify the feasibility and effectiveness of the new method.

A predicting ZWD model based on multi-source data and generalized regression neural network

LI Junyu, YAO Yibin, LIU Lilong, ZHANG Bao, HUANG Liangke, CAO Liying

2023, 52(9):  1492-1503.  doi:10.11947/j.AGCS.2023.20220084

Asbtract ( )   HTML ( )   PDF (10361KB) ( )  

References | Related Articles | Metrics

Tropospheric wet delay is a more difficult part of GNSS error sources to be corrected. Most of the approved empirical models of zenith wet delay (ZWD) are based on single-source data (i.e. radiosonde data or reanalysis data), and the variation patterns of ZWD on different scales are characterized by preset model functions, so it is difficult to accurately describe the nonlinearly complex variations of ZWD on different scales, and the accuracy needs to be further improved. To address this issue, a predicting ZWD model is constructed based on multi-source data with higher spatiotemporal resolution and a generalized regression neural network (GRNN) with strong nonlinear approximation capability. Firstly, grid ZWD of two different data sources is optimized and downsampled by a GRNN model, and merged with radiosonde ZWD to obtain high-quality ZWD dataset. Then, the input and the output vectors of the GRNN training model is designed according to the characteristics that ZWD is greatly affected by time and space and the characteristics of machine learning. Finally, a posteriori optimization method is used to determine the model parameters, and then the optimal forecasting model is obtained. Validated by the radiosonde ZWD, in comparison with the approved empirical GPT2w model and the single-source (i.e. radiosonde) data model with the same method, the accuracy of the proposed model is improved by 25.3% and 11.1% respectively in terms of RMS. And the accuracy of the proposed model has good spatiotemporal stability. In addition, the computational efficiency and PPP application experimental results show that the computational efficiency of the proposed model can meet the needs of GNSS real-time applications, and the improvement effect of PPP is better than that of GPT2w.The proposed model obtains high ZWD forecasting accuracy without setting the model function, which provides an idea for ZWD modeling.

Photogrammetry and Remote Sensing

Multi-scale building instance refinement extraction from remote sensing images by fusing with decentralized adaptive attention mechanism

JIANG Baode, HANG Wei, XU Shaofen, WU Yong

2023, 52(9):  1504-1514.  doi:10.11947/j.AGCS.2023.20220322

Asbtract ( )   HTML ( )   PDF (12602KB) ( )  

References | Related Articles | Metrics

The accurate and efficient automatic extraction of building footprints from remote sensing images has a wide range of applications. Since the buildings in remote sensing images have different types, scales, shapes and backgrounds, the existing methods, to varying degrees, suffer from the problems of missing small-scale buildings, blurred contour boundaries, and inability to distinguish individual building instances. Therefore, this paper proposed a multi-scale building instance refinement extraction convolutional neural network(MBRef-CNN) fusing with decentralized adaptive attention mechanism for remote sensing images. First, a feature pyramid network fused with split-attention and adaptive attention mechanism (SA-FPN) was used to learn multi-scale building features. Then, according to the multi-scale features, the region proposal network (RPN) was used to detect the location of individual building instances. Finally, the boundary refinement network (BndRN) was used to iteratively acquire the precise building masks. On WHU aerial imagery dataset, the comparison experiments were conducted with the existing popular segmentation methods. The results show that the accuracy of the proposed method in this paper is higher than the others. Moreover, the MBRef-CNN shows good comprehensive performance in multi-scale building extraction, and has obvious accuracy advantages in small-scale building extraction.

The automatic stitching algorithm with anti-parallax for wide-baseline weak-texture images

YAO Guobiao, HUANG Pengfei, GONG Jianya, MENG Fei, ZHANG Jin

2023, 52(9):  1515-1527.  doi:10.11947/j.AGCS.2023.20220417

Asbtract ( )   HTML ( )   PDF (21162KB) ( )  

References | Related Articles | Metrics

Based on the available algorithm, it is a tough work to achieve stitching of wide-baseline weak-texture images with parallax discontinuity. As a result, the stitching task usually requires manual intervention. For this, we modify the critical steps of image matching and image registration, and propose an anti-parallax automatic stitching algorithm for wide-baseline weak-texture images. First, we obtain the quasi-dense correspondence of weak-texture features from coarse to fine, based on the local feature transformers model incorporating the geometric correction of the image perspective. Next, based on matching points and deep neural network (DNN), the reliable perspective transform between wide-baseline images can be learned to eliminate global registration disparity, and then the local left disparities are precisely fitted by thin plate spline (TPS) function. Furthermore, the polygon boundary of the image stitching result is regularized, and it is trained as a regularized rectangle through a fully convolutional network, which effectively removes the blank area and preserves the content of the image stitching to the maximum extent. Finally, four groups of UAV and ground close-range wide-baseline stereo image pairs with weak-textures are selected and tested, and the results of image matching and registration stages of our method are respectively compared with the results of the existing representative algorithms. The experimental results verify that our method has significant advantages in the number of matching points, accuracy and stitching quality, and show good stability at the weak-texture and parallax discontinuity regions of the images.

Automated texture mapping of reality CSG building model with oblique aerial imagery

CHENG Xu, GE Liang, ZHANG Fan, HUANG Xianfeng

2023, 52(9):  1528-1537.  doi:10.11947/j.AGCS.2023.20220165

Asbtract ( )   HTML ( )   PDF (18021KB) ( )  

References | Related Articles | Metrics

Accurate geometric and texture information is required to build highly realistic building models. In building three-dimensional modeling, some appurtenances like window sills and curved balcony are simplified or neglected, which brings errors to the original geometric structure and results in large deviations and obvious misalignment in texture mapping. We propose an automated texture mapping method using oblique imagery to solve this problem. Firstly, introducing the model geometric error to constrain the texture seam lines position, which avoids passing the area with significant deviations; Next, using line structure of the building façade to construct the registration model, which directly deals with the misalignment; Then, the color adjustment is carried out. Experiments using reality CSG (constructive solid geometry) building models of various complexity were conducted to verify the effectiveness of the proposed method, which obtains textures with continuous geometric contour and consistent color.

A change detection network with joint spatial constraints and differential feature aggregation

WEI Chuntao, GONG Cheng, ZHOU Yongxu

2023, 52(9):  1538-1547.  doi:10.11947/j.AGCS.2023.20220345

Asbtract ( )   HTML ( )   PDF (7789KB) ( )  

References | Related Articles | Metrics

Change detection aims to observe the expression differences of ground objects in different time series. Deep learning has become the mainstream method to achieve this task. In the existing remote sensing change detection methods based on deep learning, they generally focus more on learning deep semantic features in images, while ignoring the semantic advantages and gaps between different levels of features resulting in insufficient detection performance. To this end, this paper proposes a change detection network that combines spatial constraints and difference feature aggregation. By controlling the flow of feature information in the network, the difference between the low-level feature and high-level semantic information of the detection object is eliminated, and the quality of prediction results is improved. Firstly, the siamese network is used in combination with the feature pyramid structure to generate multi-scale differential features; Then, the proposed coordinate self attention mechanism (CSAM) is used to constrain the low-level features, strengthen the edge structure of the change area and the accurate position information, and combine the classical convolutional attention module to fully capture the context change feature information; Finally, the gated fusion mechanism is used to extract the channel relationship and control the fusion of multi-scale features to generate a change image with clear boundary and complete interior. A large number of experiments are carried out on the change detection dataset CDD and LEVIR-CD, and compared with the existing change detection network models, the proposed method shows the best detection effect in different scenarios.

Probabilistic seismic landslide hazard mapping with consideration of slope occurrence environment

CHEN Shuai, MIAO Zelang, WU Lixin

2023, 52(9):  1548-1561.  doi:10.11947/j.AGCS.2023.20220213

Asbtract ( )   HTML ( )   PDF (31051KB) ( )  

References | Related Articles | Metrics

Displacement model is a significant method for probabilistic seismic landslide hazard assessment. However, as the main inputs of the displacement model, the physical-mechanical parameters of rock mass only consider the lithological differences, while the spatial heterogeneous of the rock mass strength resulted from the slope occurrence environment was ignored before, which affects the reliability of the seismic landslide hazard assessment. Based on analysis to co-seismic landslides resulted from Wenchuan earthquake and study on the weighting of impact factors, this paper presents a new method for probabilistic seismic landslide hazard assessment with consideration of slope occurrence environment and develops a group of ArcGIS-based modules for probabilistic seismic landslide hazard mapping. Taking Kangding as an experimental example, the reliability of the proposed method is verified by using the landslides from remote sensing visual interpretation and field survey. Finally, according to the peak ground acceleration under different earthquake scenarios provided by the fifth-generation seismic ground motion parameters zonation map of China, the permanent seismic displacements of Kangding under different earthquake scenarios are computed, and the seismic landslide hazards are mapped. The results show that under the frequently encountered earthquake scenario, some areas in Kangding are high seismic landslide hazard areas mainly affected by the tectonic environment and rainfalls; while under the rarely encountered earthquake scenario, the area seriously affected by earthquakes in Kangding increases significantly and shows an obvious spatially clustering distribution, especially in the vicinity of Kangding and where along Kangding-Luding segment of national highway G318. Besides, the uncertainty of seismic landslide hazard mapping caused by rock mass strength difference is quantified and analyzed, which shows that rock mass strength is critical for the uncertainty of assessment results. For application, suitable rock mass strengths should be selected for different scenarios to conduct probabilistic seismic landslide hazard assessment and mapping. The presented method expands the GIS application in geohazard mapping, which is of general adaptability and could be improved further. The experimental results are valuable for future land planning and utilization in Kangding, and for prevention of earthquake-induced landslide and safeguard of Sichuan-Tibet railway.

Cartography and Geoinformation

Multi-hierarchy hexagonal grid traffic model for off-road path planning

CHEN Zhanlong, WU Beibei, WANG Run, DAI Weiwei, XU Daozhu, MA Chao

2023, 52(9):  1562-1573.  doi:10.11947/j.AGCS.2023.20220159

Asbtract ( )   HTML ( )   PDF (12808KB) ( )  

References | Related Articles | Metrics

Aiming at the low efficiency of path planning in a large-scale off-road environment, this paper proposes a multi-hierarchy hexagonal grid traffic model for off-road path planning. The model can improve the execution efficiency of the path planning algorithm while reducing the grid data scale and maintaining the rationality of the planned path. Based on the hexagonal grid unit, this paper designs the quantification rule of traffic capacity gives each grid the corresponding traffic capacity, constructs the traffic model of ordinary hexagonal grid, and then establishes the multi-hierarchy grid compression rules. The adjacent grids are merged, and the adjacent grid relationship is reconstructed to generate a traffic model containing different hierarchy of grids. Finally, according to the multi-hierarchy hexagonal grid traffic model proposed in this paper, a heuristic function considering the elements of slope and ground cover is designed, and the A^* path planning algorithm is further optimized. Experiments show that the multi-hierarchy hexagonal grid traffic model proposed in this paper reduces the number of grids by 53.75% and the time required for path planning by 57% compared with the ordinary hexagonal grid traffic model.

A building selection method supported by graph convolutional network

AN Xiaoya, ZHU Yude, YAN Xiongfeng

2023, 52(9):  1574-1583.  doi:10.11947/j.AGCS.2023.20220216

Asbtract ( )   HTML ( )   PDF (13964KB) ( )  

References | Related Articles | Metrics

As a fundamental aspect of map generalization, building selection requires thorough consideration of various contextual factors such as size, orientation, shape, density, and more. However, many existing methods have only focused on single or a few factors, often relying on manual selection rules and parameters, which limits their practicality. In this study, we propose a data-driven building selection method using the graph convolutional network(GCN). Our method organizes buildings into a graph using Delaunay triangulation, with nodes representing building centers and edges denoting adjacent relationships between buildings. The size, orientation, shape, and density of each building are computed as the descriptive features for the associated nodes. Further, a GCN is constructed by stacking multiple graph Fourier convolutions and trained with semi-supervised learning to enable it to decide whether a building is selected or not. Experiments show that our method can effectively learn selection knowledge with few labels and perform well in maintaining the original spatial distribution densities and selecting important individual objects. It overcomes the difficulties in rule definition and parameter setting of traditional methods and does not rely on a large number of manual labels, which provides a promising solution for intelligent generalization.

Vector reconstruction algorithm for building footprints orientation calculation and regularization

LIU Changzhen, MA Wei, MA Hong, WEI Shixuan

2023, 52(9):  1584-1594.  doi:10.11947/j.AGCS.2023.20220082

Asbtract ( )   HTML ( )   PDF (8033KB) ( )  

References | Related Articles | Metrics

Building footprints regularization is an elementary and preliminary work of urban spatial information extraction and application. Currently, most of the existing methods are applicable for buildings with a single orientation, whereas not suitable for complex building with multi-orientations, let alone architectural complexes. In this paper, we propose a vector reconstruction based method to calculate the orientation of the building, which is then taken as a constraint to regularize the building footprint. Firstly, we design a vector reconstruction principle. We group and transform the footprints vectors, calculate the orientation, and correct the right angle to obtain the orientation vectors of the building footprints. Then, we use the orientation vector as a constraint to determine the rotation base point. Finally, the regularization of the building footprint can be achieved by adjusting the node coordinates. Compared with the main orientation method, the experimental results show that the proposed method performs better in algorithm stability and node coordinate moving distance. The error of original data can be further reduced by regularization method. Through the proposed method, the footprints orientation calculation and regularization of complex building and architectural complexes can be effectively accomplished.

Multi-pedestrian trajectory prediction method based on multi-view 3D simulation video learning

CAO Xingwen, ZHENG Hongwei, LIU Ying, WU Mengquan, WANG Lingyue, BAO Anming, CHEN Xi

2023, 52(9):  1595-1608.  doi:10.11947/j.AGCS.2023.20220239

Asbtract ( )   HTML ( )   PDF (28228KB) ( )  

References | Related Articles | Metrics

Multi-pedestrian trajectory prediction is one of the key factors in integrating urban geographic information system and intelligent transportation. To address the problems of insufficient training data, difficult labeling, and low accuracy of pedestrian trajectory prediction in multi-view scenes for existing methods, we propose a novel multi-pedestrian trajectory prediction method based on multi-view 3D simulation video learning. First, a simulation simulator is used to generate the required multi-view pedestrian trajectory annotation data. Then, we mix up the trajectory of the selected view and the adversarial trajectory by a convex combination function to generate the enhanced adversarial trajectory. Next, an advanced detection and tracking algorithm is used to encode and track pedestrian appearance information. Furthermore, the enhanced trajectory and coding information are used as the feature input of a graph attention recurrent neural network to model pedestrian interaction. Finally, the pedestrian trajectory is decoded by a position decoder to extract pedestrian motion characteristics, and multi-pedestrian trajectory prediction is completed. The ADE and FDE accuracies of our method on the ETH/UCY fixed-view dataset are 0.41 and 0.82, respectively. The ADE accuracy on the ActEV/VIRAT and Argoverse multi-view datasets is 17.74 and 65.4, and the FDE accuracy is 34.96 and 172.8.

Summary of PhD Thesis

Development of the high-precision global models for the key parameters of troposphere considering spatiotemporal factors

HUANG Liangke

2023, 52(9):  1609-1609.  doi:10.11947/j.AGCS.2023.20220164

Asbtract ( )   HTML ( )   PDF (770KB) ( )  

Related Articles | Metrics

Water vapor retrieved from ground-based GNSS and its applications in extreme weather studies

HE Qimin

2023, 52(9):  1610-1610.  doi:10.11947/j.AGCS.2023.20220187

Asbtract ( )   HTML ( )   PDF (800KB) ( )  

Related Articles | Metrics

Research of the day-boundary discontinuities in GNSS carrier phase time transfer

ZHANG Xiangbo

2023, 52(9):  1611-1611.  doi:10.11947/j.AGCS.2023.20220217

Asbtract ( )   HTML ( )   PDF (792KB) ( )  

Related Articles | Metrics

Research on key technologies of transmission line inspection form ALS point cloud

MA Weifeng

2023, 52(9):  1612-1612.  doi:10.11947/j.AGCS.2023.20220230

Asbtract ( )   HTML ( )   PDF (772KB) ( )  

Related Articles | Metrics

Research on object detection in high resolution remote sensing imagery based on convolutional neural networks

DONG Zhipeng

2023, 52(9):  1613-1613.  doi:10.11947/j.AGCS.2023.20220234

Asbtract ( )   HTML ( )   PDF (768KB) ( )  

Related Articles | Metrics

Study on GPS coordinate time series analysis of CMONOC stations and application of the annual phase-augmented clustering algorithm

WU Shuguang

2023, 52(9):  1614-1614.  doi:10.11947/j.AGCS.2023.20220238

Asbtract ( )   HTML ( )   PDF (783KB) ( )  

Related Articles | Metrics

Spatiotemporal variation of PM_2.5 in Guanzhong Basin based on satellite remote sensing

ZHANG Kainan

2023, 52(9):  1615-1615.  doi:10.11947/j.AGCS.2023.20220270

Asbtract ( )   HTML ( )   PDF (802KB) ( )  

Related Articles | Metrics

Study on retrieving sea surface rainfall intensity, wind speed and wave height using spaceborne GNSS-R technology

BU Jinwei

2023, 52(9):  1616-1616.  doi:10.11947/j.AGCS.2023.20220709

Asbtract ( )   HTML ( )   PDF (783KB) ( )  

Related Articles | Metrics