基于Savitzky-Golay方法的遥感影像融合

Remote Sensing Image Merging Based on Savitzky-Golay Method

把Savitzky-Golay滤波方法拓展至二维,提出了基于Savitzky-Golay方法的遥感影像融合算法,并通过与HSI变换、PCA变换和小波变换融合等传统融合算法的比较,证明该方法性能最优。分析Savitzky-Golay滤波算子阶数和小波变换尺度对融合影像质量的影响,发现Savitzky-Golay滤波融合过程中Savitzky-Golay滤波算子阶数是决定融合质量的关键因素。算子阶数越高,融合后影像细节信息越丰富,但光谱信息损失也越严重;算子阶数越低,融合后影像光谱信息保持能力越好,但细节信息增强能力变弱。如何根据具体的遥感影像自动确定最佳的算子阶数是下一步要解决的问题。

A new remote sensing image merging method, presented by expanding 2 dimensions Savitzky-Golay filter to 3 dimensions, was proved to have a superior ability by comparing with the traditional merging methods such as HSI transformation, PCA transformation and wavelet transformation. The results show that the order of filter operator plays an important role in the quality of merged image according to the impact of filter operator and the scale of wavelet transformation. The merged image ob- tains more detailed information with higher order of the filter operator, but loses more spectral information, and vice versa. How to obtain optimal orders of the filter operator automatically according to the specific images is the further work.