Zhixin Duan

Zhixin Duan

Auto-generated profile, based on publicly available information

Zhixin Duan

Wuhan University

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 January 2022 in Optics Express
Verified
Journal Article
Closed access
ZDZhixin Duan
Zhixin Duan
SCSensen Chu
Sensen Chu
LCLiang Cheng
Liang Cheng
CJChen Ji
Chen Ji
MLManchun Li
Manchun Li
WSWei Shen
Wei Shen
Satellite-derived bathymetry (SDB) has an extensive prospect in nearshore bathymetry for its high efficiency and low costs. Atmospheric correction and bathymetric modeling are critical processes in SDB, and examining the performance of related algorithms and models will contribute to the formulation of reliable bathymetry strategies. This study explored the effectiveness of three general atmospheric correction algorithms, namely Second Simulation of a Satellite Signal in the Solar Spectrum (6S), Atmospheric correction for OLI ‘lite’ (ACOLITE), and QUick Atmospheric Correction (QUAC), in depth retrieval from Landsat-8 and Sentinel-2A images using different SDB models over Ganquan Island and Oahu Island. The bathymetric Light Detection and Ranging (LiDAR) data was used for SDB model training and accuracy verification. The results indicated that the three atmospheric correction algorithms could provide effective corrections for SDB. For the SDB models except log-transformed band ratio model (LBR) and support vector machine (SVM), the impact of different atmospheric corrections on bathymetry was basically the same. Furthermore, we assessed the performance of six different SDB models: Lyzenga’s model (LM), generalized additive model (GAM), LBR, SVM, multilayer perceptron (MLP), and random forest (RF). The bathymetric accuracy, consistency of bathymetric maps and generalization ability were considered for the assessment. Given sufficient training data, the accuracy of the machine learning models (SVM, MLP, RF) was generally superior to that of the empirical inversion models (LM, GAM, LBR), with the root mean square error (RMSE) varied between 0.735 m to 1.177 m. MLP achieved the best accuracy and consistency. When the depth was deeper than 15 m, the bathymetry error of all the SDB models increased sharply, and LM, LBR and SVM reached the upper limit of depth retrieval capability at 20–25 m. In addition, LM and LBR were demonstrated to have better adaptability in heterogeneous environment without training data.

28 October 2022 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
HCHui Chen
Hui Chen
LCLiang Cheng
Liang Cheng
QZQizhi Zhuang
Qizhi Zhuang
Ka ZhangKa Zhang
Ka Zhang
NLNing Li
Ning Li
ZDZhixin Duan
Zhixin Duan
The use of fully supervised deep learning methods to extract buildings from remote sensing images has shown excellent performance, which requires large amounts of training data with laborious per-pixel labeling. Compared with pixelated intensive labeling, it is much easier to label data using scribbles, which only takes few seconds for one image. In this paper, a one-stage structure-aware weakly supervised network (SAWSN) for building extraction is proposed, and it learns from easily accessible scribbles rather than from densely annotated ground truth. Firstly, to solve the problem that direct training with scribble labels will lead to poor building structures, an auxiliary edge detection task is introduced to localize building edges explicitly. Secondly, a structure aware scribble extension module (SASEM) is designed to recover building structures from scribbles through effective utilization of edge features. Finally, an edge-structure-aware loss is proposed to limit the scope of the restored structure. We perform extensive experiments on three newly labeled benchmark building extraction datasets (WHU, ISPRS Potsdam, and Vaihingen). Experimental results show that our method achieved 91.72%, 92.83%, and 92.22% of F1 using the ISPRS Vaihingen, Potsdam, and WHU datasets, respectively, and outperformed the state-of-the-art scribble-based weakly supervised methods by 3.27% of IoU .

29 February 2024 in IEEE Geoscience and Remote Sensing Letters
Verified
Journal Article
Closed access
ZDZhixin Duan
Zhixin Duan
SDShengkun Dongye
Shengkun Dongye
CJChen Ji
Chen Ji
YSYueting Song
Yueting Song
JXJianbo Xiao
Jianbo Xiao
LCLiang Cheng
Liang Cheng
Global land cover heterogeneity is usually directly reflected in the background of remote sensing objects. This substantially affects the generalization ability of object detection models and is a challenge for remote sensing object detection in large-scale areas. In this study, we analyze the heterogeneity of land cover in different climate areas, and propose a remote sensing object image synthesis method based on multi-stage blending strategy (MBS) to achieve a high-quality blending of the labeled runway foreground image and the typical surface image of the area to be detected. Three geometric augmentation algorithms are applied to expand the size of the sample data. Finally, two representative object detection models, YOLOv5 and Faster R-CNN, are used to evaluate the effectiveness of the synthetic data. Experiments in North Africa and Central Asia demonstrate that the MBS method effectively improves the performance of airport runway detection in large-scale heterogeneous areas. Additionally, our method performs better than Copy-Paste and Poisson blending.

1 January 2026 in IEEE Transactions on Geoscience and Remote Sensing
Verified
Journal Article
Closed access
TWTianjiao Wang
Tianjiao Wang
QMQingzhou Mao
Qingzhou Mao
ZDZhixin Duan
Zhixin Duan
XHXuran He
Xuran He

10 November 2025 in IEEE Sensors Journal
Verified
Journal Article
Closed access
JYJunfan Yi
Junfan Yi
WGWanxuan Geng
Wanxuan Geng
ZDZhixin Duan
Zhixin Duan
CJChen Ji
Chen Ji
KSKeke Shang
Keke Shang
LCLiang Cheng
Liang Cheng
Frequent intrusions of Unmanned aerial vehicles (UAVs) highlight the importance of UAV 3D tracking for security and surveillance applications. Vision-based systems for UAV perception have been widely adopted in recent years, primarily focusing on UAV 2D detection. However, these systems fall short in providing accurate 3D tracking, which is crucial for effective anti-UAV operations. To address this gap, we propose a method for UAV 3D tracking based on the fusion of vision and positioning sensors. The proposed two-stage method relies solely on the positions of cameras, eliminating the need for camera attitude information, which is typically required in traditional methods. In the 2D detection stage, UAVs are detected from monocular video using the You Only Look Once version 12 (YOLOv12) model, which generates bounding boxes for each UAV. Based on these bounding boxes, the GrabCut algorithm is applied to segment the UAV contour and compute its 2D centroid position. In the 3D tracking stage, we estimate the relative camera poses using 2D UAV’s position across views and combine them with GNSS-based camera positions to reconstruct the UAV’s 3D trajectory in the world coordinate system. In a 130 × 80 × 40 m field experiment, the proposed method achieved a median error of 4.41 m in 3D tracking, outperforming the 4.93 m error of the traditional Resection-Intersection method.