Search

Your search keyword '"Xie, DongHai"' showing total 33 results
Start Over Author "Xie, DongHai" Publication Year Range Last 10 years
33 results on '"Xie, DongHai"'

1. S2Looking: A Satellite Side-Looking Dataset for Building Change Detection

Author

Shen, Li, Lu, Yao, Chen, Hao, Wei, Hao, Xie, Donghai, Yue, Jiabao, Chen, Rui, Lv, Shouye, and Jiang, Bitao

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence, Electrical Engineering and Systems Science - Image and Video Processing
Abstract

Building-change detection underpins many important applications, especially in the military and crisis-management domains. Recent methods used for change detection have shifted towards deep learning, which depends on the quality of its training data. The assembly of large-scale annotated satellite imagery datasets is therefore essential for global building-change surveillance. Existing datasets almost exclusively offer near-nadir viewing angles. This limits the range of changes that can be detected. By offering larger observation ranges, the scroll imaging mode of optical satellites presents an opportunity to overcome this restriction. This paper therefore introduces S2Looking, a building-change-detection dataset that contains large-scale side-looking satellite images captured at various off-nadir angles. The dataset consists of 5000 bitemporal image pairs of rural areas and more than 65,920 annotated instances of changes throughout the world. The dataset can be used to train deep-learning-based change-detection algorithms. It expands upon existing datasets by providing (1) larger viewing angles; (2) large illumination variances; and (3) the added complexity of rural images. To facilitate {the} use of the dataset, a benchmark task has been established, and preliminary tests suggest that deep-learning algorithms find the dataset significantly more challenging than the closest-competing near-nadir dataset, LEVIR-CD+. S2Looking may therefore promote important advances in existing building-change-detection algorithms. The dataset is available at https://github.com/S2Looking/.

Published
2021
Full Text
View/download PDF

4. Mercator Descriptor: A Novel Global Descriptor for Loop Closure Detection in LiDAR SLAM

Author

Wang, Zhibo, Xie, Donghai, Wu, Yu, Wu, Haiyang, Qi, Xingyu, Huang, Daoping, Fu, Yibo, and Zhong, Ruofei

Abstract

Loop closure detection is a fundamental technique in simultaneous localization and mapping (SLAM), playing a crucial role in the map-building process. Inspired by the Mercator projection and depth maps, this article presents a novel loop closure detection method for LiDAR SLAM, termed the Mercator descriptor. Our approach applies the Mercator projection principles to the point cloud data captured by the sensor in each frame and extracts depth information to generate descriptors. Additionally, we utilize a two-step identification strategy, from coarse to fine, to improve the accuracy and recall of loop closure detection. Testing on public datasets demonstrates that our algorithm achieves an average maximum $F1$ score of 94.57%, performing well across various datasets. Notably, we have integrated our algorithm into existing SLAM frameworks, and results from multiple datasets show that our algorithm can perform real-time and efficient map optimization. To share our findings and contribute to the community, we have open-sourced our code on GitHub: https://github.com/wangzika/Mercator-Descriptor.git.

Published
2024
Full Text
View/download PDF

5. An Indoor-Outdoor Layered HD Map Construction for Unmanned Ground Vehicles.

Author

Gao, Huimin, Zhong, Ruofei, Qi, Xingyu, Xie, Donghai, Sun, Zhenxing, and Sun, Haili

Abstract

High-Definition Maps (HD Maps) are lane-level 3-dimensional road network models that ensure the safe operation of autonomous vehicles and drive the continuous advancement of autonomous driving technology to higher levels. However, HD Maps currently face challenges regarding limited coverage and description of indoor environments (e.g. underground parking), as well as the lack of uniformity between indoor and outdoor map formats. Therefore, this paper proposes a Layered HD Map model for both indoor and outdoor environments. The map model consists of 6 layers, allowing effective representation of indoor and outdoor environments, as well as transitional areas. The paper also establishes an Unmanned Ground Vehicle (UGV) to collect and update map data using LiDAR (Light Detection and Ranging) SLAM (Simultaneous Localization and Mapping) and autonomous exploration technology, aiming to enhance the efficiency and automation of HD Map acquisition. In order to verify the validity of the proposed map model, this paper conducted a navigation experiment in a double decker parking lot containing entrances and exits. The results of the experiments demonstrate that the Layered HD Map generated through autonomous LiDAR SLAM can effectively describe the indoor and outdoor environments, and enable successful navigation of UGVs. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

9. Synthesis and Detection Algorithms for Oblique Stripe Noise of Space-Borne Remote Sensing Images

Author

Li, Binbo, Xie, Donghai, Wu, Yu, Zheng, Lijuan, Xu, Chongbin, Zhou, Ying, Fu, Yibo, Wang, Chenglong, Liu, Bin, and Zuo, Xin

Abstract

Oblique stripe noise widely appears in remote sensing images after image correction, exhibiting arbitrary tilt angles and parallel distribution. Because of its arbitrary randomness in tilt angles and lengths, oblique stripe noise increases the difficulty of detection compared to vertical or horizontal stripe noise. For the first time, we propose a group of oblique stripe noise synthesis and detection algorithms combining imaging mechanisms and deep learning. To get controllable synthetic oblique stripe noise data for the training detection model, two sample augmentation methods are presented by the image correction’s imaging mechanisms with new linear transformation and the generative adversarial network (GAN) algorithm with Cycle-GAN, respectively. A large-scale simulated stripe noise dataset simulated oblique stripe noise dataset (SOSD) is simulated using these two methods. A new deep learning detection algorithm, robust detection of oblique stripe noise (RDOS), is presented considering the presence of oblique stripe noise. RDOS is trained using both SOSD and a real stripe noise dataset, and it obtains the optimal detection model for testing. The experimental results show that the accuracy reaches 82.93%, the recall rate reaches 85.17%, the $F1$ score reaches 84.04%, the average precision (AP) reaches 82.34%, and the frames per second (FPS) reaches 33.33. Compared with the general line detection models, our model exceeds ~300% in accuracy and ~60% in speed. In the future, the proposed algorithms will have great potential for application in various areas, such as quality evaluation, image preprocessing, and engineering problems related to multiangle linear object augmentation and detection.

Published
2024
Full Text
View/download PDF

18. Relative Pose Estimation and Accuracy Verification of Spherical Panoramic Image

Author

XIE Donghai, ZHONG Ruofei, WU Yu, FU Han, HUANG Xiaochuan, and SUN Zhenxing

Subjects
spherical panoramic images, essential matrix, Computer Science::Computer Vision and Pattern Recognition, relative pose estimation, epipolar line, lcsh:Mathematical geography. Cartography, SVD, lcsh:GA1-1776
Abstract

This paper improves the method of the traditional 5-point relative pose estimation algorithm, and proposes a relative pose estimation algorithm which is suitable for spherical panoramic images. The algorithm firstly computes the essential matrix, then decomposes the essential matrix to obtain the rotation matrix and the translation vector using SVD, and finally the reconstructed three-dimensional points are used to eliminate the error solution. The innovation of the algorithm lies the derivation of panorama epipolar formula and the use of the spherical distance from the point to the epipolar plane as the error term for the spherical panorama co-planarity function. The simulation experiment shows that when the random noise of the image feature points is within the range of pixel, the error of the three Euler angles is about 0.1°, and the error between the relative translational displacement and the simulated value is about 1.5°. The result of the experiment using the data obtained by the vehicle panorama camera and the POS shows that:the error of the roll angle and pitch angle can be within 0.2°, the error of the heading angle can be within 0.4°, and the error between the relative translational displacement and the POS can be within 2°. The result of our relative pose estimation algorithm is used to generate the spherical panoramic epipolar images, then we extract the key points between the spherical panoramic images and calculate the errors in the column direction. The result shows that the errors is less than 1 pixel.

Published
2017

19. Numerical investigation on rules of fracture propagation during hydraulic fracturing in heterogeneous coal-rock mass

Author

Yuan, Zhigang, Shao, Yaohua, Xie, Donghai, Huang, Fei, Yuan, Zhigang, Shao, Yaohua, Xie, Donghai, and Huang, Fei

Abstract

To investigate rules of fracture propagation during hydraulic fracturing in heterogeneous coal-rock mass, a new mathematical model for hydraulic fracturing with seepage-damage coupling and its numerical algorithm are proposed. The rules of coal-rock mass heterogeneity, confining pressure, beforehand hydraulic slotting, and non-symmetric pressure gradient on fracture propagation are investigated. Numerical results show the following: (1) Compared to homogeneous coal-rock mass, the fracture propagation pattern exhibits a more zig-zag characteristic and the fracture initiation pressure is reduced in heterogeneous coal-rock mass. (2) Fracture propagation during borehole fracturing is mainly controlled by confining pressure ratio, and the fracture would propagate along the path with least resistance in coal-rock mass. (3) During hydraulic fracturing with beforehand hydraulic slotting, fracture propagation pattern would become more complex with slotting length increasing; the propagation direction of fracture is primarily controlled by principal stress difference, the larger of principal stress difference, the more difficult of oriented fracturing. (4) Non-symmetric pressure gradient can reduce breakdown pressure and influence fracture propagation pattern, which provides a beneficial guide for oriented fracturing. The simulation results are consistent with the theoretical solutions and experimental observations, which is promising to guide field operation of hydraulic fracturing to improve coalbed methane extraction.

Published
2019

33. The development of local ambient air quality standards: A case study of Hainan Province, China.

Author

Song Q, Zhang N, Zhang Y, Yin D, Hao J, Wang S, Li S, Xu W, Yan W, Meng X, Xu X, Wu X, Xie D, Zhu Y, Qu Q, Hou X, Jiang Y, Dong Z, Zheng H, Sun Y, Li Z, and Zhao B

Abstract

The ambient air quality standard (AAQS) is a vital policy instrument for protecting the environment and human health. Hainan Province is at the forefront of China's efforts to protect its ecological environment, with an official goal to achieve world-leading air quality by 2035. However, neither the national AAQS nor the World Health Organization guideline offers sufficient guidance for improving air quality in Hainan because Hainan has well met the former while the latter is excessively stringent. Consequently, the establishment of Hainan's local AAQS becomes imperative. Nonetheless, research regarding the development of local AAQS is scarce, especially in comparatively more polluted countries such as China. The relatively high background values and significant interannual fluctuations in air pollutant concentrations in Hainan present challenges in the development of local AAQS. Our research proposes a world-class local AAQS of Hainan Province by reviewing the AAQS in major countries or regions worldwide, analyzing the influence of different statistical forms, and carefully evaluating the attainability of the standard. In the proposed AAQS, the annual mean concentration limit for PM 2.5 , the annual 95th percentile of daily maximum 8-h mean (MDA8) concentration limit for O 3 , and the peak season concentration limit for O 3 are set at 10, 120, and 85 μg/m 3 , respectively. Our study indicates that, with effective control policies, Hainan is projected to achieve compliance with the new standard by 2035. The implementation of the local AAQS is estimated to avoid 1,526 (1,253-1,789) and 259 (132-501) premature deaths attributable to long-term exposure to PM 2.5 and O 3 in Hainan in 2035, respectively., Competing Interests: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper., (© 2023 The Author(s).)

Published
2023
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results