High-precision localization of the Mars rovers is fundamental for path planning and safe navigation towards exploration targets during Mars missions. In cross-site visual localization, image matching is the key step to obtain corresponding points connecting images from different sites. The cross-site visual localization method based on Affine SIFT used in Tianwen-1 mission but is constrained in regions of Mars with poor texture and large viewpoint invariance. In this paper, we propose a cross-site visual localization methodology of Mars rover based on self-supervised keypoint extraction and robust matching. The self-supervised keypoint extraction network, which is called MRSS-Net, employs multi-scale deformable structures (MSDS) during the feature encoding stage to enhance the network’s ability of extracting invariant features in regions with large viewpoint variations and improve the rate of identical points for cross-site images with poor texture. In addition, we develop self-attention descriptor enhancement mechanism (SADEM) to distinguish local features in repetitive patterns. The robust matching, which is called adaptive 2D-3D matching, employs GNC dead reckoning (3D priori information) to construct the initial coarse matching domain and homography matrix (2D information) to construct a progressively shrinking refined matching domain. We compared our method against Affine SIFT based cross-site visual localization model, as well as advanced deep learning algorithms and evaluate the performance using NaTeCam images collected during the traversal of four long-distance traversals (a total of 44 Martian sol sites) by Zhurong rover. The experimental results show that our framework reduces the localization error by 12.5% and improves localization robustness by 50.8%, compared to Affine SIFT based cross-site visual localizaztion method used in Zhurong rover. Additionally, our method outperforms state-of-the-art deep-learning techniques and ensures the current accuracy of cross-site visual localization for Mars rover, while significantly increasing the level of automation.

Lunar orbiter image matching is a critical process for achieving high-precision lunar mapping, positioning, and navigation. However, with the Moon’s weak-texture surface and rugged terrain, lunar orbiter images generally suffer from inconsistent lighting conditions and exhibit varying degrees of non-linear intensity distortion, which pose significant challenges to image traditional matching. This paper presents a robust feature matching method based on crater neighborhood structure, which is particularly robust to changes in illumination. The method integrates deep-learning based crater detection, Crater Neighborhood Structure features (CNSFs) construction, CNSF similarity-based matching, and outlier removal. To evaluate the effectiveness of the proposed method, we created an evaluation dataset, comprising Multi-illumination Lunar Orbiter Images (MiLOIs) from different latitudes (a total of 321 image pairs). And comparative experiments have been conducted using the proposed method and state-of-the-art image matching methods. The experimental results indicate that the proposed approach exhibits greater robustness and accuracy against variations in illumination.

High-precision localization of Mars rovers is essential for safe path planning and efficient navigation toward scientific targets. As planetary rovers traverse the surface, their positional uncertainty accumulates, which can be corrected through global localization by registering rover images to orbital maps. To date, image-based solutions are widely adopted; however, substantial manual intervention is often required, which is time-consuming and limits the range of autonomous navigation. To address these challenges, we propose a two-stage localization framework, comprising the Mars cross-view few-shot training paradigm (MarsCVFP), Mars cross-view feature extraction network (MCVN) trained under MarsCVFP, and a robust template matching algorithm. Specifically, the MarsCVFP model can leverage implicit cross-view feature as guidance without relying on a large amount of high-precision location-level supervision and explicitly annotated, specific learning targets in the scene. MCVN can capture discriminative fine-grained features on the weakly textured and unstructured surface of Mars by constructing the multi-scale feature pyramid structure (MSFPS) and the feature interaction module (FIM). We validate our framework on 85 unit-planned sites and 20 panoramic sites, respectively, as traversed by the Zhurong rover. The experimental results demonstrate that our framework consistently outperforms both the traditional approaches and the representative learning-based methods across diverse terrains, including dunes, bedrock, craters, and flat plains, achieving a localization success rate above 82% while maintaining a localization accuracy of better than 4 pixels, even under coarse prior positions uncertainties spanning 40 m × 40 m.

Lunar optical and synthetic aperture radar (SAR) imagery provide complementary information for characterizing the lunar surface. However, their joint use remains challenging because of substantial cross-modality differences and severe illumination constraints, particularly in polar regions. To address this challenge, we propose LS2ODiff (Lunar SAR-to-Optical Diffusion), a diffusion-based framework designed for SAR-to-optical image translation in lunar environments. LS2ODiff uses SAR observations as conditional guidance in the diffusion process and incorporates a partial-convolution strategy into the U-Net backbone to handle irregular invalid regions. In addition, self-attention modules are incorporated into the downsampling stages of the U-Net to model long-range spatial dependencies and enhance global structural consistency in complex lunar terrain. We further construct a dedicated paired dataset of the lunar south polar region by registering Chandrayaan-II DFSAR data with Lunar Reconnaissance Orbiter (LRO) Narrow-Angle Camera (NAC) imagery. Comparative experiments against Pix2Pix, CycleGAN, SynDiff, and ConDiff demonstrate that LS2ODiff achieves better visual fidelity and quantitative performance in terms of peak signal-to-noise ratio (PSNR), structural similarity index measure (SSIM), Fréchet inception distance (FID), and learned perceptual image patch similarity (LPIPS). These results demonstrate the potential of diffusion models for high-fidelity lunar image translation, offering new opportunities for polar terrain interpretation and future exploration missions.