A remote sensing image pan-sharpening method based on spectral-spatial balanced adaptive model and polar lights optimizer with two-dimensional ising model in spherical coordinate system

Abstract Pansharpening enhances remote sensing imagery by fusing PAN and MS data to combine spatial detail with spectral information, making it one of the key techniques for improving image quality and interpretability. A spectral-spatial balanced adaptive pansharpening method was proposed based on the two-dimensional Ising Model Polar Lights Optimizer (2D-IPLO) constructed in spherical coordinates. Firstly, an adaptive injection model with dynamically regulated spectral-spatial consistency is developed, in which spectral correlation weights are introduced to achieve an adaptive balance between spectral fidelity and spatial detail enhancement. Secondly, oscillation factors derived from the physical mechanism of the 2D Ising model are designed and embedded into the key dynamic weights of both the global and local search phases of the PLO, enabling a synergistic interplay between the two search strategies. This design significantly accelerates convergence and enhances optimization performance. In the CEC2022 benchmark tests, 2D-IPLO demonstrates superior convergence speed and overall optimization capability. When integrated into the proposed adaptive framework and evaluated on multiple satellite datasets, both qualitative and quantitative results confirm that the proposed method delivers outstanding performance and exhibits strong application potential.