Dual-LiDAR systems are widely deployed in autonomous driving, yet extrinsic calibration remains challenging in non-overlapping field-of-view (FoV) configurations where correspondence-based methods are unreliable. We propose an engineering-oriented 2.5D calibration framework that estimates horizontal extrinsics $(x,y,\mathrm{yaw})$ via motion-guided planar alignment and then refines them using Gaussian Process Implicit Surfaces (GPIS), which provide continuous and probabilistic surface constraints from spatially disjoint scans. This design avoids calibration targets and reduces dependence on strong scene assumptions, improving robustness under noise and weak structure. Extensive high-fidelity simulation experiments demonstrate centimeter-level lateral accuracy and sub-degree yaw error, consistently outperforming representative motion-based and BEV-based baselines under both clean and noisy settings. To further assess real-world applicability, we conduct a preliminary nuScenes case study by splitting LiDAR scans into front and rear subsets to emulate a non-overlapping dual-LiDAR setup, achieving improved yaw accuracy and competitive lateral precision. Overall, the proposed method serves as a practical refinement stage for non-overlapping dual-LiDAR calibration, with a favorable balance of accuracy, robustness, and engineering feasibility.

High-resolution satellite imagery supports scalable mapping, but dense-to-vector pipelines can misrank overlapping candidates and retain local geometric errors. We present SAGE-Map, a framework for structure–appearance guided calibration and geometry refinement built on our SAM-Road-based pipeline for OpenSatMap20. The baseline extracts Lane, Curb, and Virtual polylines with normal-displacement correction. SAGE-Map combines structural scoring with candidate-aligned DINOv2 evidence to rescore Lane candidates, then trims unsupported endpoints while preserving candidate identity, class, and count. On 393 validation tiles, SAGE-Map achieves 29.27, 31.33, 34.06, and 36.06% Chamfer average precision at 0.9, 1.5, 3.0, and 4.5 m, respectively, and 46.78% mean intersection over union (mIoU). Under our reconstructed implementation of the published protocol, these point estimates are higher than reported UniMapGen aggregates; no paired statistical comparison with UniMapGen is available. Paired bootstrap intervals for gains over our baseline are positive through 3.0 m but cross zero at 4.5 m. Post hoc component comparisons associate most of the strict-tolerance gain with rescoring and the mIoU gain with endpoint trimming. Additional geometric analysis shows lower forward distance but worse symmetric distance and reduced GT length coverage after trimming. The gains therefore concern the evaluated ranking and overlap metrics, rather than general geometric completeness.