基于NSCT的遥感图像融合性能评价及分析

Performance evaluation and analysis of remote sensing image fusion based on non-subsampled contourlet transform

为了寻求比小波变换更加有效的图像多分辨率分析方法,提出了一种基于非采样Contourlet变换(NSCT)和区域特性选择的遥感图像融合方法,并对NSCT在遥感图像融合中的性能及计算复杂性进行了深入分析。通过与离散小波变换(DWT)和采样的Contourlet变换(CT)的对比融合实验,结果表明,基于NSCT的融合方法计算复杂性较DWT明显提高,取得的融合效果更好。

Two-dimension wavelet is isotropic and has poor directional selectivity. It can only reflect characteristic and location of the singularity, but hardly express the high-dimensional geometrical feature such as edges and textures of. images. So as to Two-dimension wavelet has some limitation in using of remote sensing image fusion, and to find out the more effective multi-resolution analysis method is necessary. A novel fusion strategy used the region-based selection operator based on non-subsampled contourlet transform is presented, and do a thorough research on the performance appraisal and computational complexity analysis of remote sensing image fusion based on non-subsampled contourlet transform, at the same time. In the experiments we compared this strategy with the DWT and CT. The experimental results show that the computational complexity of the fusion based on NSCT obviously improve than DWT, but it can achieve the best fusion performance.