Global mapping of lunar surface chemistry is crucial for revealing the geological characteristics and evolutionary history of the Moon. However, existing estimates of elemental abundances rely primarily on remote sensing data calibrated with sample-based ground truth information from the lunar nearside, leaving the farside largely unconstrained and limiting the accuracy of global chemical models. Here we integrate farside ground truth data from the Chang’e-6 sampling site, together with pre-existing nearside sample data and orbital spectral datasets from the Kaguya multiband imager, to refine global chemical maps. We apply a residual convolutional neural network with a fine-tuning strategy to optimally calibrate elemental abundances across the surface. The resulting global maps constrain the extent and composition of farside terranes and reveal deep-seated materials exposed in the South Pole–Aitken basin and highlands. These refined maps offer quantitative guidance for landing site selection and future lunar exploration missions. Farside ground truth information gathered by Chang’e-6, integrated with orbital and nearside sample data, refines global lunar chemical maps and provides a framework to guide future exploration.

Accurate registration between optical and synthetic aperture radar (SAR) imagery remains challenging, particularly in rugged terrain, where SAR-specific geometric distortions significantly degrade correspondence reliability. Existing optical-SAR registration methods primarily focus on mitigating radiometric discrepancies across sensors, while the terrain-dependent failure mechanisms induced by SAR imaging geometry have received limited attention. In this article, we reformulate optical-SAR image registration as a geometry-consistency correction problem across heterogeneous imaging models, rather than a direct multimodal matching task. A digital surface model (DSM) is introduced as an explicit geometric proxy that bridges optical and SAR projections, transforming geometric distortions from sources of interference into usable constraints for correspondence establishment. Based on this formulation, a terrain-aware two-stage registration framework is developed. In the first stage, reliable correspondences are established in geometry-dominant rugged regions through DSM-mediated projection consistency. In the second stage, these correspondences are propagated to appearance-dominant flat regions using a distortion-aware multimodal matching strategy. A unified block adjustment model integrates correspondences from different terrain regimes to achieve globally consistent coregistration. Extensive experiments on 1108 image pairs and multiple challenging rugged scenarios demonstrate that the proposed method consistently outperforms representative approaches in terms of registration accuracy and robustness, particularly under severe geometric distortions.

Objective With the advancement of lunar exploration missions, autonomous lander navigation has emerged as a critical technology for ensuring mission safety. Craters serve as essential navigation landmarks, enabling precise lander positioning in scenarios lacking alternative navigation infrastructure or during signal interruptions. However, rapid and stable high-precision crater detection faces significant challenges due to illumination variations, resolution discrepancies, and noise interference. Traditional image processing methods exhibit high computational complexity and struggle to meet real-time requirements. Deep learning approaches achieve higher detection rates but depend on extensive datasets and powerful computing resources. Conversely, lightweight machine learning methods demonstrate insufficient generalization capability under complex lighting and diverse terrain conditions. This is particularly evident as crater detection accuracy and center extraction precision degrade substantially with changes in solar elevation and azimuth angles. Methods To achieve fast and accurate crater detection in lunar orbital images, this paper proposes a method combining principal component analysis (PCA) feature representation with linear discriminant analysis (LDA) hashing. The template image undergoes grayscale vectorization to eliminate the influence of overall brightness. PCA is then applied for dimensionality reduction, extracting principal component features robust to illumination and resolution changes. Subsequently, supervised learning with LDA Hashing generates a hash mapping function, compressing these features into binary hash codes. A crater template library, covering diverse illumination conditions, is then constructed. During the detection stage, an image pyramid structure is introduced to handle craters of varying scales. This is achieved by generating multi-resolution image layers through downsampling. A sliding window technique, combined with varying strides and window sizes, then performs scans at each layer. For every detection window, PCA features are extracted. Recognizing the critical influence of illumination on feature extraction, a primary illumination direction correction module is integrated. This module intelligently selects illumination-consistent crater samples from the template library for matching based on the sun azimuth or imaging time. The extracted PCA feature vector of the detection window is mapped to a hash code using the pre-trained hash function. The Hamming distance between this hash code and those of the template craters is computed in Hamming space to identify potential crater locations. Finally, the normalized cross-correlation (NCC) algorithm is employed for precise verification. It calculates the NCC coefficient between candidate regions and template images, effectively eliminating false detections. Non-maximum suppression is then applied to fuse detection results across different scales, yielding the definitive crater locations. Results and Discussions Experiments meticulously investigate the impact of illumination on crater detection, leading to crucial insights for the construction of robust template libraries. Results (Figs. 4 and 5) demonstrate that template library robustness requires adherence to specific construction principles: solar elevation angles sampled at 10 degrees intervals from 0 degrees (horizon) to 90 degrees (zenith) to cover terminator-to-noon scenarios; solar azimuth angles sampled at 20 degrees intervals from 0 degrees to 360 degrees , generating 180 azimuthal templates for comprehensive directional adaptability. Validation leverages imagery from the lunar reconnaissance orbiter camera (LROC) wide angle camera (WAC) and high-resolution descent cameras from the Chang'e-4 (CE-4) and Chang'e-6 (CE-6) missions. Quantitative analysis confirms the significant contribution of the sliding window strategy and illumination correction to multi-scale detection. Comparative evaluations with mainstream feature detection algorithms reveal superior performance in both accuracy and efficiency (Table 1). Specifically, on the CE-6 image, the total detection rate (TDR), correct detection rate (CDR), and detection rate (DR) reach 0.91, 0.89, and 0.82 respectively. For the CE-4 image, the results are 0.85 (TDR), 0.89 (CDR), 0.77 (DR). The LROC WAC image achieves a TDR of 0.91, a CDR of 0.94, and a DR of 0.86. Computational times are remarkably low at 0.06 s for CE-6, 0.07 s for CE-4, and 0.05 s for LROC WAC. These computation times significantly outperform those of traditional template matching and deep learning methods. This efficiency is attributed to the low-dimensional principal component feature representation, which effectively mitigates background clutter and noise interference around craters. Ablation studies further confirm the critical role of the illumination correction module and NCC verification. Removing illumination correction causes notable declines in CDR and DR. Eliminating NCC verification maintains a high TDR but drastically reduced CDR, with CE-6 DR dropping to 0.66. This is because, while hashing enables rapid screening in low-dimensional space, it remains susceptible to edge artifacts, high-frequency noise, and complex backgrounds (Fig. 7). These vulnerabilities can lead to duplicate detections and false positives. Fortunately, NCC verification effectively resolves these problems. Conclusions In summary, the proposed method effectively addresses the challenges of multi-scale detection, complex illumination, and real-time requirements for crater detection in lunar orbital images. This provides reliable technical support for lunar geological research and lander navigation. However, the method may still exhibit some missed detections for craters with blurred edges. Future work will focus on integrating region fusion strategies or edge enhancement mechanisms. This should further improve recognition robustness in complex backgrounds, thereby expanding its practical application potential in lunar autonomous navigation and precise positioning.

ABSTRACT With the resurgence of lunar exploration, long‐range rover traverses in challenging regions such as the Lunar South Pole demand high‐precision autonomous absolute localization, as rugged terrain and extreme illumination limit current ground‐in‐the‐loop methods. Our study proposed a non‐parametric absolute localization approach that used lunar impact craters as landmarks to support high‐precision rover positioning during extended traverses. We developed two crater detection methods to extract craters' spatial information from rover's stereo imagery, i.e., a 3D point cloud‐based method using stereo matching network, and a deep learning‐based method that combined 2D keypoints from object detection with disparity maps. A novel object‐level dual‐constraint crater similarity metric was developed and integrated into the particle filter localization framework for efficient crater spatial information matching against the orbital reference database, thus enabling absolute rover localization. To validate the approach, we built a high‐fidelity simulation environment in Blender, using real topographic data from the Intuitive Machine 1 (IM‐1) Nova‐C Odysseus landing region near the Lunar South Pole with varying illumination condition settings (including different solar elevation and azimuth angles). Extensive experiments demonstrated that the approach achieved accurate and robust localization, with an average absolute error under 1.5 m across a 1058.65‐m traverse. These results showed the potential of the proposed approach for precise lunar rover remote localization and navigation in challenging environments.

Lunar landing optical navigation and positioning is one of the key technologies of the lunar exploration mission. Landing images may exhibit low contrast, leading to weak internal texture features, and may also present significant radiometric discrepancies relative to the reference images. Traditional feature-matching methods are prone to mismatching, making it difficult to accurately determine the position of the lander. In this paper, a block-optimized FDAFT (BOFDAFT, Block-Optimized Fast Double-Channel Aggregated Feature Transform) feature matching algorithm is proposed, which is combined with a position estimation algorithm for lander position estimation. Additionally, a novel metric, referred to as Distribution Uniformity of Matched Points (DUMP), is proposed to quantify the spatial distribution uniformity of the matched points. By comparing with other matching methods, it is verified that the proposed matching algorithm is feasible in a variety of complex scenarios, and the method is validated by using simulation data and Chang'E-6 landing images. The average ratio of the magnitude of position deviation to the lander's altitude is 0.6%, and the average reprojection absolute error in trajectory recovery of Chang'E-6 landing images is 1.77 pixels. This study has demonstrated the effectiveness of the proposed method for navigation and position in complex environments, which can provide technical support for future lunar exploration missions.

Abstract Lunar crater-based visual navigation has become one of the key methods for lunar spacecraft navigation. In such navigation systems, the accuracy of the crater’s elevation modelling is a key determinant of positioning reliability. However, existing research primarily focuses on optimizing crater matching algorithms, with limited attention given to the impact of elevation errors and their transmission pathways on navigation accuracy. This paper addresses this gap by systematically analysing the mechanisms of elevation error generation and quantifying their effects on positioning and reprojection errors through the geometric constraints of the Perspective-n-Point (PnP) algorithm. Simulation results demonstrate a significant linear increase in both positioning and reprojection errors as crater elevation errors grow. Specifically, when the elevation error reaches 500 meters, the mean positioning error increases to approximately 960 meters. This study offers valuable insights into the precision requirements for elevation modelling in lunar navigation systems, which are essential for optimizing system design and enhancing positioning accuracy.

Despite humanity's many lunar missions to the equatorial and mid-latitude regions, the south pole remains uncharted because of its exceptionally harsh conditions. The quest for water ice and the drive to establish lunar bases have positioned the south pole area above 80° latitude, characterized by permanently shaded regions and conducive to water ice preservation. However, the daunting terrain and intricate illumination in this area present significant challenges to engineering safety. Here, we introduce a Landing Feasibility Probability (LFP) model to evaluate the viability of potential landing sites. We pinpoint 120 prospective landing sites, stratified into 25 high-priority, 64 medium-priority, and 31 low-priority sites. These sites, encompassing a mere 0.6 % of the lunar south pole, have been rigorously vetted against 10 critical factors for landing feasibility. These sites show a pronounced clustering around 8 major craters and plateaus, organizing into 7 lunar site networks, with each network comprising 4 sites with a maximum dimension under 25 km, ideal for the development of lunar bases and observational networks. The LFP model's selection process is derived from a heuristic Genetic Algorithm (GA) informed by expert experience. The landing sites are strategically positioned to address the dual challenges of scientific goals and engineering safety at the lunar south pole, while also providing site selection guidance and facilitating international collaboration and communication for lunar expeditions. This method can also be adapted for site selection on other celestial bodies (e.g. Mars and asteroids) for scientific exploration and construction of extraterrestrial bases.

Accurate modeling of solar irradiance on Mars is essential for future exploration missions, including landing site selection, rover energy optimization, and habitat planning. Existing models, however, often rely on broad meteorological approximations and neglect the complex interplay between atmospheric scattering, terrain occlusion, and solar geometry, particularly in areas with complex topography. This paper introduces a novel, high-resolution solar irradiance modeling method that integrates detailed terrain and atmospheric data, constructing high-fidelity horizon profiles and calculate full radiative transfer to accurately model both direct and diffuse solar radiation components. This approach not only enhances the accuracy of solar irradiance simulations on the Martian surface but also generates high-precision spatiotemporal irradiance maps, enabling detailed illumination analysis over localized areas, along specific trajectories, or across extended timeframes. The model was rigorously validated using 30 sols of data (sols 35-64) from the Mars Environmental Dynamics Analyzer aboard the Perseverance rover, achieving a root mean square error (RMSE) of 6.57 W m-2 and a mean absolute percentage error (MAPE) of 4.07%. Notably, in areas with significant terrain variation during periods of low solar elevation, the model reduced the MAPE by up to 6.7% and decreased the RMSE by up to 9.7 W m-2 compared to atmosphere-only models, demonstrating a substantial improvement in accuracy. Furthermore, the model accurately predicted abrupt irradiance changes due to terrain-induced solar obstruction, a phenomenon not captured by traditional methods. These results underscore the model's significance in providing critical, location-specific irradiance data and demonstrate its potential to inform optimal path planning for rovers, site selection for future habitats, and a deeper understanding of the Martian environment.

Landing point localization is of great significance to the lunar exploration engineering and scientific missions. Vision-based landing point localization methods have successfully been utilized in Chang'e series missions. The issues in landing point visual localization task containing low-resolution reference maps, illumination changes between descent images and maps, and low automation of the localization workflow still need to be solved. In this article, a high-precision and automatic landing point visual localization method with high-resolution map generation is proposed, including initial localization of the first frame, hybrid fine matching, landing point propagation in descent sequence images, and absolute position estimation for landing point. High-resolution digital elevation model and digital orthophoto map (DOM) are generated from Lunar Reconnaissance Orbiter Camera Narrow Angle Camera images and SLDEM2015 data. Phase-based image matching method is adopted for initial localization and matching between descent image and reference map to enhance the illumination robustness. The performance of our method is validated using the descent sequence images from Chang'e series missions. For Chang'e-6 lander, the estimated landing point coordinates are (-153.9870 degrees +/- 0.00002 degrees, -41.6378 degrees +/- 0.00001 degrees, -5256.962 m +/- 0.0041 m). Compared with the manually measured lander coordinates, the deviation of landing point position is less than 1 pixel in DOM.

Objective High-precision optical autonomous navigation is a key technology for deep-space exploration and planetary landing missions. During the descent phase of a lunar lander, communication delays and accumulated errors in the Inertial Navigation System (INS) lead to significant positioning deviations, which pose serious risks to safe landing. Optical images acquired by the lander are matched with pre-stored lunar landmark databases to establish correspondences between image coordinates and three-dimensional coordinates of lunar surface features, thereby enabling precise position estimation. This process is challenged by dynamic illumination variation on the lunar surface, noise in prior pose information, and limited onboard computational resources. Traditional template matching methods exhibit high computational cost and sensitivity to rotation and scale variation. Keypoint-based methods, such as Scale Invariant Feature Transform (SIFT) and Speeded Up Robust Features (SURF), suffer from uneven keypoint distribution and sensitivity to illumination variation, which results in reduced robustness. Deep learning-based approaches, including SuperPoint, SuperGlue, and LF-Net, improve feature detection accuracy but require substantial computational resources, which restricts real-time onboard deployment. To address these limitations, a landmark matching algorithm is proposed that integrates gray-gradient dual-channel features with deformation parameter optimization, enabling high-precision and realtime matching for lunar optical autonomous navigation. Methods Dual-channel image features are constructed by combining gray-level intensity and gradient magnitude representations. Gradient features are computed using Sobel operators in the horizontal and vertical directions, and the gradient magnitude is calculated as the Euclidean norm of the two components. To reduce the effect of local brightness variation and ensure inter-region comparability, zero-mean normalization is applied independently to each feature channel. An adaptive weighting strategy is employed, in which weights are assigned according to local gradient saliency, and a bias term is introduced to retain weak texture information, thereby improving robustness under noisy conditions. Landmark matching is formulated as a nonlinear least-squares optimization problem. A deformation parameter vector is defined, which includes incremental rotation, scale, and translation relative to the prior pose. The objective function minimizes the weighted sum of squared differences between dual-channel landmark features and image features, with Tikhonov regularization applied to constrain parameter magnitude and improve numerical stability. The Levenberg-Marquardt (LM) algorithm is adopted to iteratively estimate the optimal deformation parameters. Its adaptive damping strategy enables switching between gradient descent and Gauss-Newton updates, ensuring stable convergence under large prior pose errors. Iteration terminates when the error norm falls below a predefined threshold or when the maximum iteration number is reached, yielding the optimal landmark transformation parameters. Results and Discussions Experiments are conducted using simulated lunar landing images generated from 60 m-resolution SLDEM (Digital Elevation Model Coregistered with SELENE Data)data, with high-fidelity illumination rendering applied to ensure realistic lighting conditions (Fig. 2). To evaluate matching performance under different scenarios, 143 landmarks are synthesized with systematically controlled perturbations in rotation, scale, and translation. Four representative methods are selected for comparison, including convolution-accelerated Normalized Cross-Correlation (NCC), SURF-based feature matching with image enhancement, globally and locally optimized NCC, and the proposed algorithm ( Fig. 4). The results indicate clear performance differences among the methods. Convolution-accelerated NCC achieves sub-second runtime and demonstrates high computational efficiency, although its accuracy degrades under gray-level variation and geometric deformation, with mean absolute errors of 2.41 px along the x-axis and 3.37 px along the y-axis, and a success rate of 89.51% (Table 1). SURF-based matching achieves sub-pixel accuracy, with mean absolute errors of 0.56 px along the x-axis and 0.54 px along the y-axis, although its success rate is limited to 48.95% and its runtime exceeds one second, which restricts onboard applicability. The globally and locally optimized NCC method exhibits the lowest accuracy, with errors of 4.54 px along the x-axis and 4.92 px along the y-axis, and the longest runtime of 4.41 s, despite achieving a 100% success rate. In contrast, the proposed algorithm consistently achieves sub-pixel accuracy comparable to SURF, maintains a 100% success rate, and sustains a stable runtime of approximately 0.5 s across all test cases. Its robustness to landmark deformation and illumination variation demonstrates suitability for complex operational conditions. Overall, the results show that the proposed algorithm achieves a favorable balance among accuracy, robustness, and computational efficiency. Conclusions A landmark matching algorithm is presented that integrates gray-gradient dual-channel features with deformation parameter optimization. Gray-level intensity and gradient magnitude information from both landmark templates and lander images are jointly exploited to construct a dual-channel matching model that minimizes feature differences. Deformation parameters, including rotation, scale, and translation, are iteratively optimized using the LM algorithm, enabling rapid estimation of the optimal landmark position in the lander image. Experimental results show stable convergence within sub-second runtime, with an average matching error of 1.03 pixels under disturbances in attitude, scale, and position. The proposed method outperforms single-channel gray-level cross-correlation and SURF-based matching approaches in accuracy, robustness, and efficiency. These results provide practical support for the design and implementation of future autonomous lunar optical navigation systems.

Identifying minerals on Mars is crucial for finding evidence of water on the planet. Currently, spectral inversion methods based on remote sensing data are primarily used; however, they only provide sparse and scattered maps of mineral exposures. To address this limitation, we propose a multi-scale spatial association modeling framework (MSAM) that couples the geographical distribution of Martian hydrous minerals with environmental factors based on the existence of spatial dependence, to achieve dense and continuous mapping of hydrous minerals. Our approach leverages explanatory variables - such as elevation, slope, and aspect - to establish spatial associations with potential areas of hydrous minerals, selected via multiscale search ranges from existing hydrous mineral exposures. These association results are used to identify potential hydrous mineral locations and estimate probabilities for potential hydrous mineral points. High-probability points are then combined with known exposures, and Kriging interpolation is applied to produce a continuous surface map. Finally, the interpolation results are evaluated using geomorphological maps, along with correlation analysis. The proposed MSAM enhances prediction accuracy and addresses the challenges of incomplete detection and undetected areas inherent in remote sensing-based spectral inversion. Results reveal that incorporating environmental factors reduces the RMSE by 25% and improves spatial correlation by 30% compared to traditional interpolation techniques. An overlay analysis intersecting the interpolated results with geomorphologic features obtained through semantic segmentation further demonstrates a coupling relationship between hydrous minerals and geomorphologic features within a specific spatial range.

In lunar exploration, high-resolution topography is an important basis for safe landing and mission planning. Remote sensing images are the main data sources for the reconstruction of lunar surface topography [1]. Among them, the orbiter images preserve the topographic photometric information under different illumination directions, and the descent images contain high-resolution morphological details of the landing site. In order to integrate the advantages of multi-illumination directions of orbiter images and high resolution of descent images, we propose a joint photometric-constrained method for topography reconstruction using both orbiter and descent images. In the framework of the joint photometric-constrained Shape from Shading (SfS) [2-4], the photometric information in multi-source images illuminated from different directions is added into the cost function as a weighted regular term in topography reconstruction. We focus on the Chang'E-3 landing site. We used the Lunar Reconnaissance Orbiter (LRO) Narrow Angle Camera (NAC) images of the area and Chang'E-3 descent images for experiments, and obtained topographic data of the site with a resolution better than 0.1 m/pixel. Comparing with previously derived topography [5], we verified that our topography is more consistent result with the images in multi-angle illumination rendering [6], integrating the photometric information of the multi-source images and preserving the morphological details such as small-size impact craters. The method proposed in this study not only improves the accuracy of topography reconstruction of the Chang'E-3 landing site, but also provides a new idea for the joint processing of multi-source image data.[1] Di K., et al. (2020) Topographic mapping of the moon in the 21st century: from hectometer to millimeter scales. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, XLIII-B3-2020, pp.1117-1124.[2] Horn, B.K.P. (1990) Height and gradient from shading. International Journal of Computer Vision, 5, pp. 37–75.[3] Beyer R.A., et al. (2018) The Ames Stereo Pipeline: NASA's Open Source Software for Deriving and Processing Terrain Data. Earth and Space Science, 5, pp. 537-548.[4] Tenthoff M. et al. (2020) High Resolution Digital Terrain Models of Mercury. Remote Sensing, 12, p. 3989.[5] Henriksen M.R., et al. (2017) Extracting accurate and precise topography from LROC narrow angle camera stereo observations. Icarus, 283, pp.122-137.[6] Tong X., et al. (2023) A high-precision horizon-based illumination modeling method for the lunar surface using pyramidal LOLA data. Icarus, 390, p. 115302.

Digital Surface Models (DSMs) extracted from multi-view satellite images have extensive applications in the filed of photogrammetry. Although Neural Radiance Fields (NeRF) has shown significant potential in 3D reconstruction, most existing NeRF methods for satellite scenes adopt an end-to-end single-branch network structure, making it difficult to achieve fine-grained modeling of multiple complex terrain simultaneously, and performs poorly in weak-texture regions. Meanwhile, deep MLP structures are prone to information degradation during feature transmission, further affecting the completeness and accuracy of DSMs. To address these challenges, we propose RDS-NeRF, a novel NeRF framework integrating residual feature enhancement and depth supervision. The method introduces a residual feature enhancement structure to alleviate the problem of information degradation during feature transmission in the network and improve the model’s ability to model local details and low-texture regions. Additionally, estimated depth maps are incorporated as global geometric priors to guide the network in constructing more accurate and complete 3D structures. Experiments on the WorldView-3 satellite imagery datasets across multiple typical land cover types (building, road, water body, and vegetation) and complex scenes integrating multiple land features demonstrate that RDS-NeRF outperforms mainstream methods in terms of DSM accuracy, completeness, and novel view synthesis quality. Ablation experiments further validate the complementarity and effectiveness of the residual enhancement and depth supervision mechanisms across different scene types. In conclusion, RDS-NeRF provides a new and effective solution for generating high-quality DSMs from satellite imagery with adaptability to multiple scenes. Code will be available at https://github.com/dfsvdgf/RDS-NeRF.

During the mosaicking of orthophotos, geometric and radiometric inconsistencies between adjacent images can cause misalignments at the boundaries, necessitating seamline detection to bypass prominent ground features. Existing methods struggle to simultaneously circumvent various ground features, such as buildings, ridges, and farmland in large-scale remote sensing images with rich ground features. This leads to insufficient robustness in densely built urban areas and mountainous village regions. To address this issue, this article proposes a seamline extraction method based on an adaptive cost A* algorithm. Initially, the seamline network is extracted as a distance-cost constraint to limit the search range of the algorithm, ensuring efficiency. Then, an improved A* algorithm is employed to optimize the seamline, utilizing an adaptive obstacle threshold and structural similarity index to construct the heuristic function. This approach effectively bypasses prominent ground features in both urban and mountainous areas. The method does not require pre-extraction of image features and constructing a cost map; instead, it dynamically detects impassable regions in the overlapping areas during the A* path search process. Finally, the image is reprojected and mosaicked based on the seamline network. This method does not rely on auxiliary data and is more suitable for large-scale image datasets with extensive coverage. Experimental results demonstrate that, compared to existing methods and commercial software, the proposed method shows superior radiometric and geometric consistency, overcoming the robustness issues of current large-scale remote sensing image mosaicking methods.

Integrating optical satellite images and laser altimetry data is essential to achieve high-accuracy mapping, especially in areas without ground control points. However, traditional methods suitable for relatively stable regions have limitations in Antarctica due to the dynamic characteristics of changing ice sheet surfaces. This article proposes an optimized multicriteria method for selecting laser elevation control points and image tie points. This method facilitates robust block bundle adjustment of Chinese ZY-3 satellite images, using elevation constraints from NASA's ICESat altimetry data. It includes filtering of laser altimetry points based on multiple criteria to ensure stable and reliable elevation control/check points, interstrip image tie points filtering based on ice flow velocity, and an iterative weighting strategy to mitigate negative effects from potential mismatches. Four comparative experiments are designed and conducted using two strips of ZY-3 images covering an area of about 77 km x 465 km in East Antarctica. The results show that the original elevation positioning error of the ZY-3 stereo images is 10.62 m, but it increases significantly to 65.35 m when using the unfiltered laser points as control. With laser point selection but without interstrip tie points filtering, the elevation stereo positioning accuracy can only be improved limited to 5.26 m, and it can be significantly further improved to 2.60 m when tie points filtering is applied, which has validated the necessity and effectiveness of the proposed method.

Conventional pixel-level seamline detection algorithms exhibit exponential time complexity on large, batch-mode remote-sensing mosaics, making it difficult to achieve an optimal trade-off between accuracy and efficiency. This article introduces a globally optimal and highly efficient seamline detection framework. First, a preliminary seamline network is generated by iteratively clipping valid orthoimage regions with a Voronoi diagram, and image blocks are extracted only within overlap areas to markedly reduce data volume. Second, a cost graph constructed on down-sampled blocks is traversed in a reverse-diagonal Z-pattern; a "local entropy-gradient" composite cost function is applied, and a linear-time dynamic-programming (DP) scheme rapidly produces coarse seamlines that bypass texture-rich regions and confine the search space to a narrow band. Third, a buffer centered on the coarse seamline is created, within which an enhanced Dijkstra algorithm performs pixel-level refinement to accurately avoid complex obstacles. Experiments on the GF-7 data set demonstrate that, compared with five representative methods-SMP-DP, A*, Dijkstra, graph-cut, and OrthoVista-the proposed approach improves geometric accuracy by 14.46%, 58.69%, 50.20%, 17.79%, and 69.30%, respectively; processing efficiency is increased by 12.74%, 19.19%, 49.89%, >500%, and 83.72%, respectively. The algorithm has successfully mosaicked 627 GF-7 scenes covering the entire Henan Province, and has yielded similarly favorable results on ZY-3, GF-1, and GF-3 imagery, underscoring its high applicability and robustness for multisource, large-format remote-sensing production.

Accurate deformation measurement of building materials has consistently been a focal point of research within the field of material testing. This paper proposes a robust 3D crack parameter estimation approach to measure the crack propagation characteristics of concrete pillars under compression testing. Through the use of an improved multiple-window matching strategy and other advanced image processing algorithms, the precise positional information of speckle target points can be calculated in speckle image sequences. 3D point cloud data and full-field deformation of target points on the concrete pillar surface can be further calculated through photogrammetric analysis and spatiotemporal analysis, respectively. Finally, a robust crack estimation algorithm based on grid geometry analysis is proposed to extract accurate crack parameters in the presence of complex speckle texture interference. To verify the superiority and reliability of the proposed approach, both a simulation test and an empirical test were conducted. The experimental results, including positional comparisons with previous matching strategies and displacement comparisons with third-party equipment, corroborate the effectiveness of our method.

The Lunar Orbiter Laser Altimeter (LOLA) is currently the most precise spaceborne laser altimeter in lunar exploration. The digital elevation model (DEM) generated from the laser points obtained by LOLA is also the highest-precision global lunar terrain dataset available to date, and is widely used as foundational data in lunar research. However, given the uncertainties in orbit determination and the deviations in laser pointing, some of the laser profiles exhibit geolocation errors. Furthermore, due to the inherent characteristics of the polar orbits of spacecraft, the gaps between laser profiles gradually increase as the latitude transitions from polar to mid-low latitude regions. These factors collectively lead to artifacts and data voids in the original LOLA DEM. In this paper, to address these issues, we first propose a batch self-constrained adjustment method that separates tracks with significant anomalies for individual adjustment based on their slopes and evaluates the point density of each track to select the appropriate adjustment strategy. For the gaps between laser profiles, the Selenological and Engineering Explorer (SELENE)/Lunar Reconnaissance Orbiter (LRO) digital elevation model (SLDEM) is introduced as an additional data source. In this process, a co-registration method is used to reduce the inconsistencies between these two datasets. The effectiveness of the method proposed in this paper was verified in both polar and mid-low latitude regions. Finally, a high-precision terrain map of the Chang’e-6 sampling area located in the South Pole-Aitken basin was generated using the designed block processing scheme.