Guozheng Xu

Guozheng Xu

Auto-generated profile, based on publicly available information

Guozheng Xu

School of Remote Sensing and Information Engineering, Wuhan University, Wuhan, China

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

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

Representative Publications

View All Publications
3 July 2024 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
Li LiLi Li
Li Li
GXGuozheng Xu
Guozheng Xu
XZXinyan Zhou
Xinyan Zhou
Jian YaoJian Yao
Jian Yao
In the burgeoning remote sensing image data era, the swift and precise retrieval of images from extensive databases has emerged as a critical challenge. This need is particularly pronounced in applications such as environmental and disaster monitoring, resource investigation, and ground target monitoring, all heavily reliant on remote sensing images. The efficacy of image retrieval hinges significantly on advanced feature extraction methods. However, remote sensing images often suffer from disturbances caused by rich and complex backgrounds. How to extract key regions from remote sensing images, reduce background interference, and improve retrieval accuracy has become a hot research topic. Addressing this challenge, in this paper, we propose a region of interest (ROI) guided attention network (RGAN) designed to detect key category regions of targets. This network integrates a Class Activation Map (CAM) module into a deep learning framework for image retrieval. Firstly, the CAM identifies multiple categories corresponding to different remote sensing categories. Secondly, multiple category features are fed into an ROI-attention module to distinguish the importance of the category. The attention module highlights the category to be detected by suppressing interference from the background. Finally, two branches, the globally extracted image features and the locally extracted features of important categories obtained through the attention module, are integrated to form a comprehensive image representation optimized for retrieving the target object. The efficacy of our proposed method is validated through experiments conducted on diverse datasets, demonstrating an improvement in retrieval accuracy.

9 June 2025 in IEEE Geoscience and Remote Sensing Letters
Verified
Journal Article
Closed access
GXGuozheng Xu
Guozheng Xu
SYSiyuan You
Siyuan You
ke liuke liu
ke liu
Li LiLi Li
Li Li
Jian YaoJian Yao
Jian Yao
Building roof plane segmentation is important for three-dimensional (3D) building model reconstruction from airborne light detection and ranging (LiDAR) point data. During the roof plane segmentation, challenges such as pseudo planes, over- and under-segmentation often arise, particularly evident in the boundary regions. To improve the accuracy of plane segmentation, various energy optimization-based methods have been proposed to refine the roof planes. However, the existing methods optimize the energy function at the point level, which may lead to getting stuck in local optima. To address these problems, we propose a coarse-to-fine boundary relabeling approach for roof plane segmentation. Starting from an initial plane segmentation result, the proposed method iteratively refines the planes by adjusting the boundaries from the voxel level to the point level. In addition, we also design a new energy function that considers accurateness, smoothness and compactness to guide the optimization. The experimental results constructed on two datasets demonstrate that the proposed method outperforms the existing roof plane segmentation methods, achieving high accuracy and smooth boundary extraction. The source code of the proposed approach will be publicly available at https://github.com/Li-Li-Whu/Coarse2FineRoofPlane.

11 December 2023 in The Photogrammetric Record
Verified
Journal Article
Closed access
YGYingwei Ge
Yingwei Ge
BGBingxuan Guo
Bingxuan Guo
GXGuozheng Xu
Guozheng Xu
YLYawen Liu
Yawen Liu
XJXiao Jiang
Xiao Jiang
ZPZhe Peng
Zhe Peng

12 November 2024 in IEEE Geoscience and Remote Sensing Letters
Verified
Journal Article
Closed access
HYHongche Yin
Hongche Yin
PZPengwei Zhou
Pengwei Zhou
GXGuozheng Xu
Guozheng Xu
GHGaoming He
Gaoming He
Li LiLi Li
Li Li
Jian YaoJian Yao
Jian Yao
The global optimization-based color correction approach aims to minimize the color differences of multiple images by optimizing the correction model for each image. The color differences in multisource and multitemporal remote sensing images are difficult to express using a simple correction model with few parameters. When employing a more flexible correction model, the number of correction parameters and optimization equations grows rapidly with the increase in the number and resolution of input images. In addition, the correction parameters of all images are coupled together and need to be solved simultaneously. An excessive number of parameters results in solving slowly or potential failure. To solve this problem, we propose a parallelizable color correction approach that decouples the correlation of correction parameters in the optimization equations and optimizes each image separately. First, we introduce auxiliary variables that replace values related to other images in the cost function. Second, we construct optimization equations for each image and parallelly solve the correction parameters. Finally, we correct the input images through a weighted correction model to better eliminate correction artifacts. Our approach iteratively optimizes auxiliary variables and correction parameters until the correction results converge. The experimental results on several challenging datasets show that our approach significantly improves execution efficiency and obtains the global optimal solution using the flexible correction model.

30 July 2025 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
HYHongche Yin
Hongche Yin
YLYinxuan Li
Yinxuan Li
PZPengwei Zhou
Pengwei Zhou
GXGuozheng Xu
Guozheng Xu
Jian YaoJian Yao
Jian Yao
Li LiLi Li
Li Li
Orthoimage mosaicking aims to make the seamline bypass obvious objects to produce seamless digital orthophoto maps. Nevertheless, stitching artifacts may still occur when obvious objects are inevitably crossed by the seamline or when the orthoimages are not accurately aligned in geometry. Image warping is necessary to correct the geometric misalignments. Existing image warping approaches primarily focus on natural images, with relatively few studies targeting orthoimages. Unlike natural images, orthoimages have significant geometric attributes and no longer satisfy the epipolar constraint. Therefore, most existing image warping methods are ineffective in dealing with geometric misalignment in orthoimage mosaicking. To solve the above problems, we propose a coarse-to-fine image warping approach using trust region optimization for orthoimage mosaicking. Firstly, we obtain an optimal seamline and search for geometrically misaligned regions along the seamline. Secondly, for each region, we model the geometric alignment problem as an optimization problem of the deformation vectors for pixels. Next, the trust region method is employed to iteratively solve the optimization problem. We introduced the image pyramid strategy to achieve efficient coarse-to-fine optimization and help to avoid local optima. Finally, the orthoimages are warped based on the optimized pixel deformation vectors. However, when dealing with misaligned regions covered by multiple images, the frame-to-frame warping cannot achieve the optimal result. Therefore, we propose a multi-image joint optimization strategy that introduces auxiliary variables to fuse information from multiple orthoimages, ensuring seamless mosaicking. Experimental results demonstrate that whether in two-orthoimage or multi-orthoimage mosaicking, our solution has better visual and quantitative performance.

31 July 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
GXGuozheng Xu
Guozheng Xu
AZAnnan Zhou
Annan Zhou
PZPengwei Zhou
Pengwei Zhou
CXChunxia Xiao
Chunxia Xiao
Jian YaoJian Yao
Jian Yao
LLLi Li
Li Li
In recent years, deformable PatchMatch-based methods have performed well in Multi-View Stereo (MVS) due to their ability to adaptively propagate plane hypotheses from reliable pixels to unreliable ones. However, they also suffer from depth quality degradation in challenging regions, such as large textureless areas and highly reflective surfaces, because these methods still rely completely on image feature similarity. To address these problems, we propose a novel Monocular Depth-Guided Edge-Enhanced Deformation MVS (MED-MVS) method that deeply integrates monocular depth priors generated by MoGe-2 to improve the performance of deformable PatchMatch. Firstly, to avoid the unexpected edge-skipping when constructing deformable patches, we leverage the geometric edge information of the monocular normal map to constrain patch deformation within continuous areas. Then, to improve the robustness of matching cost between two patches, we design a new multi-cue cost function that combines photometric and monocular relative depth consistency. This cost is applied to evaluate the confidence of hypotheses. Finally, to further correct the depth of challenging regions and generate a depth map with sharp edges, we propose a novel monocular depth-guided filtering strategy to refine the depth map. Comprehensive experiments conducted on ETH3D and Tanks and Temples datasets demonstrate that the proposed MED-MVS can achieve state-of-the-art performance, and show a notable improvement in reconstruction completeness while maintaining high overall depth accuracy.