一种自适应的基于局部小波系数特征的遥感图像融合方法

An Adaptive Remote Sensing Images Fusion Method Based on Local Feature of Wavelet Transform

光谱保持和高分辨率保留是影像融合的两个重要问题。本文提出了一种自适应的基于局部小波系数特征的遥感影像融合方法。该方法在对多光谱影像进行IHS变换的基础上,对多光谱的I分量和高分辨率的全色影像分别进行小波多分辨率分析,而后对分解得到的近似分量以及各层各方向的细节分量利用移动模板逐一提取对应的小波系数矩阵的局部特征,采用本文提出的自适应融合准则在小波域进行影像融合,最后通过小波逆变换得到新的I′分量,与H,S分量一起还原到RGB空间,最终得到融合后的高分辨率多光谱彩色图像。本文采用一组TM多光谱图像和SPOT全色图像数据进行融合实验,利用标准差、熵,光谱扭曲度等5个重要评价指标对融合效果进行数理分析。其实验融合图像的目视效果和统计指标均优于IHS融和方法和小波融合方法。

he Spectral characteristics preservation and the high spatial resolution retention are two key issues in image fusion. A new adaptive remote sensing image fusion algorithm based on traditional IHS transform and wavelet transform was proposed. The original I element of IHS and SPOT image were decomposed into multi-resolution representation by using wavelet transform. The approximatied images and detail images resulting from wavelet transform were infused in each level of different input resources based on local coefficients feature. This paper proposed fusion rules aiming at the approximated images and detail images respectively. The resultant image can be obtained by performing inverse wavelet transform and inverse IHS transform. We used 5 indices to assess the new method comparing with the usual IHS＋ Wavelet fusion method. A set of TM band images and a SPOT image were used in our test, the experimental results show that the effect of our method is more satisfactory.