【摘要】 月面地形信息对于嫦娥3号的安全降落是至关重要的。本文提出了一种基于压缩感知的超分辨率DEM重建方法,得到了虹湾(嫦娥3号的拟着陆位置)的超分辨率DEM。该方法先根据经过模糊处理并加入噪声的低分辨率DEM重建原始的高分辨率DEM,采用K-SVD算法完成高、低分辨率过完备字典Ah和Al的学习;再获得低分辨率DEM块的稀疏表示,并将表示系数用于高分辨率字典以生成对应的高分辨率DEM块;最后运用最小二乘算法得到满足重构约束的高分辨率DEM。实验验证了算法的有效性,表明其在视觉效果及RMSE指标上均优于插值方法。更多还原

【Abstract】 The high-resolution topographic information is crucial to the achievement of scientific goals in 2nd stage of Chang’E project.In particular,large scale landing site mapping will be extremely important for the 2013 Chang’E-3 landing mission.Therefore,a super-resolution reconstruction algorithm via compressed sensing was presented to produce super resolution DEM of Sinus Iridum,the Chang’E-3 landing site in the paper.The target was to reconstruct original DEM from its blurred and down-scaled noisy version.The K-SVD algorithm was used in training a pair of low-resolution and high-resolution dictionaries.The representation coefficients,getting from the sparse representation of the low-resolution DEM about the low-resolution dictionary Al,were applied to the corresponding high-resolution dictionary Ah to reconstruct high-resolution DEM.At the end,the high-resolution DEM which satisfied the reconstruction constraint was achieved by using Least Squares algorithm.Numerical experiments demonstrated the effectiveness of the proposed algorithm.Moreover,the proposed algorithm outperforms interpolation based method in terms of visual quality and Root Mean Squared Error(RMSE).更多还原