基于NSCT与IHS变换的SAR与多光谱图像融合

Fusion of SAR and Multispectral Image Based on NSCT and IHS Transform

针对SAR图像和多光谱图像提出了一种基于NSCT与IHS变换的图像融合方法。首先对多光谱图像进行IHS变换;然后对SAR图像和多光谱图像的I分量分别进行NSCT分解,在不同频域子带系数选择时,针对高频系数和低频系数分别采用不同准则进行融合得到新的I分量;最后经IHS逆变换得到融合图像。为验证算法性能,分别选用2组不同空间分辨率比的SAR与多光谱图像开展融合实验,采用信息熵、平均梯度、相关系数等客观指标与主观评价相结合的方式,对融合结果进行分析。结果表明,该方法优于传统融合方法。

According to difference of SAR image and multispectral image, a new method based on the nonsubsampled Contourlet transform and IHS transform was put forward in this paper. Firstly, IHS transform was performed on the multispectral image, and then the intensity component of SAR image and multispectral image were decomposed by NSCT. For coefficients in different frequency domain, high frequency coefficients and low frequency coefficients adopted with different criteria respectively were fused to get the new intensity component. Finally, the fusion image was obtained by inverse IHS transform. To validate the performance of algorithm, two groups of the different spatial resolution ratio of SAR images and multispectral images, had carried out to experiment the fusion experiments. A combination of subjective evaluation and objective indicators such as information entropy, average gradient, correlation coefficient, was used to analyze the fusion results of the different methods. The experimental results show that the method is superior to the traditional fusion method like IHS transform, wavelet transform-IHS wavelet and NSCT transform.