Zhenglong Cai

Zhenglong Cai

Auto-generated profile, based on publicly available information

Zhenglong Cai

Beijing University of Civil Engineering and Architecture, The School of Geomatics and Urban Spatial Informatics, Beijing, China

Research Focus & Metrics

Academic Career Length

3Years (2024-Present)

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

Representative Publications

View All Publications
17 March 2025 in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Verified
Journal Article
Open access
JYJunxing Yang
Junxing Yang
ZCZhenglong Cai
Zhenglong Cai
TWTianjiao Wang
Tianjiao Wang
TYTong Ye
Tong Ye
HGHaoran Gao
Haoran Gao
HHHe Huang
He Huang
Ture Digital orthophoto maps (TDOMs) have become essential sources of spatial geographic data, characterized by their high precision, detailed representation, and accessibility. However, the generation of orthophotos using image differential correction techniques, which utilize digital elevation models (DEM) and digital surface models (DSM), often results in significant non-orthogonal distortions. These distortions arise from insufficient building information and the inherent limitations of DSM accuracy. Conventional techniques for geometric stitching and uniform radiometric correction are limited in their ability to enhance orthophoto quality. Although advancements in radiance-field-based view synthesis techniques are notable, they face challenges related to training speed, rendering quality, and scene editability. Even cutting-edge technologies, such as Instant-NGP, struggle to balance orthophoto quality with computational efficiency. In response to these challenges, this paper introduces Ortho-3DGS, an innovative methodology for generating DOMs from UAV imagery via 3D Gaussian Splatting. Unlike methods based on Neural Radiance Fields (NeRF), our approach employs three-dimensional Gaussian ellipsoids to represent scenes, continuously refining Gaussian parameters through depth supervision and gradient optimization for explicit scene reconstruction. The proposed methodology includes an orthophoto rendering pipeline specifically designed to produce high-quality, distortion-free DOMs. Experimental findings demonstrate that Ortho-3DGS outperforms traditional algorithms, such as ContextCapture, Pix4Dmapper, and Metashape, as well as NeRF-based methods like Ortho-NeRF and Ortho-NGP, in both radiometric quality and geometric accuracy. By utilizing explicit scene representation, depth-supervised optimization, and an efficient orthophoto rendering pipeline, Ortho-3DGS effectively reconstructs geometric and color information, enabling rapid orthophoto generation. Furthermore, our method achieves geometric accuracy comparable to leading commercial photogrammetry software and state-of-the-art NeRF-based orthophoto algorithms, while effectively addressing distortions, biases, and artifacts in complex scenarios. These advancements establish Ortho-3DGS as a robust tool for high-precision orthophoto generation and a wide range of geospatial applications.

1 October 2025 in Journal of Applied Remote Sensing
Verified
Journal Article
Closed access
ZCZhenglong Cai
Zhenglong Cai
HHHe Huang
He Huang
TYTong Ye
Tong Ye
WBWenyao Bian
Wenyao Bian
TWTianjiao Wang
Tianjiao Wang
JYJunxing Yang
Junxing Yang
To address geometric distortion, texture artifacts, and low reconstruction efficiency in orthophoto generation from sparse views, we propose a structure-consistent 3D Gaussian splatting method. By incorporating matching priors, the method enforces 3D structural consistency, reducing surface discontinuities and voids. A differentiable orthographic projection is designed to generate orthophotos from sparse inputs. Experiments over a 10 km2 area using only 36 images (191 m altitude, 4 m focal length) show that our approach substantially improves geometric completeness and texture fidelity compared with traditional methods and unmodified 3DGS. The method achieves efficiency gains of 31.5% over 3DGS and 6.9% to 9.1% over commercial software (ContextCapture, MetaShape, Pix4D). Under sparse-view conditions, it attains an average absolute error of ∼0.22 m and an average relative error of 0.52%, comparable to traditional software, demonstrating reliable geometric precision and robustness for high-quality orthophoto generation with limited image coverage.

1 October 2024 in Journal of Applied Remote Sensing
Verified
Journal Article
Closed access
PDF file
TWTianjiao Wang
Tianjiao Wang
JYJunxing Yang
Junxing Yang
ZCZhenglong Cai
Zhenglong Cai
XCXiuyuan Chi
Xiuyuan Chi
HHHe Huang
He Huang
FDFei Deng
Fei Deng
True orthophotos have become one of the most important spatial foundational geographic information due to their accurate geometric information and realistic texture features. The traditional method of producing true orthophotos is limited by the accuracy of digital elevation model (DEM)/digital surface model (DSM)/three-dimensional (3D) models. It inevitably has non-orthophoto issues, such as facade distortions and edge deformations, especially in areas with buildings. To address these challenges, we propose an innovative method that leverages a neural radiance field integrated with multi-resolution hash encoding. This method generates orthophotos directly from multi-view images and does not require additional data such as DEM, DSM, or 3D models. In comparison with the existing methods, our experimental results have achieved high-quality orthophotos, addressing various issues such as building facade distortions, building edge deformations, tree area deformations, and the presence of moving objects in orthophotos. In addition, evaluating the orthophoto generation method requires a ground truth dataset. We offer an open-source dataset containing multi-view images of unmanned aerial vehicle scenes and corresponding ground truth orthophotos of the area. This dataset comprises real-world and synthetic data, encompassing diverse scenes such as cities, trees, lakes, and more. It can be used for a comprehensive assessment of the quality of orthophotos produced by different methods. Our proposed method has been thoroughly evaluated using a wide range of challenging datasets. The experimental results demonstrate that our method outperforms traditional algorithms and the relatively latest method.

1 July 2026 in Journal of Electronic Imaging
Verified
Journal Article
Closed access
JYJunxing Yang
Junxing Yang
HGHaoran Gao
Haoran Gao
CHChunyu Huang
Chunyu Huang
XLXingchen Liu
Xingchen Liu
ZCZhenglong Cai
Zhenglong Cai
HHHe Huang
He Huang
In outdoor scene reconstruction, dynamic occlusions and multiscale structures often undermine multiview consistency and hinder effective gradient accumulation of high-frequency Gaussian primitives, leading to artifacts and the loss of fine details in Gaussian splatting–based radiance field methods. To address these challenges, we propose an uncertainty-aware hierarchical Gaussian splatting framework for outdoor 3D reconstruction. Specifically, our method constructs a hierarchical octree-based spatial representation from the results of aerial triangulation. It introduces level of detail constraints to enable structured management and progressive optimization of Gaussian primitives across different scales. This design effectively alleviates the imbalance in training and the redundant growth of Gaussian primitives commonly observed in multiscale outdoor scenes. In addition, we incorporate an uncertainty prediction mechanism that evaluates the consistency between rendered results and ground-truth images in the feature space, allowing the model to automatically identify dynamically occluded regions and suppress their gradient contributions during optimization. As a result, the adverse impact of dynamic artifacts on static scene modeling is substantially reduced. Experimental results demonstrate that, without incurring significant additional training overhead, our method consistently improves structural consistency and fine-detail reconstruction quality in outdoor scenes, while simultaneously reducing model complexity and maintaining real-time rendering performance. Furthermore, the proposed approach can be seamlessly integrated into multiple mainstream Gaussian splatting frameworks, exhibiting strong robustness and promising potential for practical deployment.