基于Curvelet变换的SAR与TM图像融合研究

An Algorithm for Image Fusion Based on Curvelet Transform

Curvelet变换作为一种具有各向异性特征的多尺度变换理论,克服了小波变换难以表达图像边缘方向特性等内在的缺陷。将Curvelet变换应用于图像融合中,能够更好地提取原始图像的特征,为融合图像提供更多的信息。文中利用Curvelet变换对同一场景的不同传感器获得的合成孔径雷达(SAR)图像和专题绘图仪(TM)图像进行融合,并对融合结果进行了客观和主观分析,实验结果表明,相比于传统的基于小波变换的图像融合算法,该算法具有更好的融合效果。

We present an algorithm based on curvelet transform for fusing SAR （Synthetic Aperture Radar） image and TM （Thematic Mapper） image. We first employ IHS （Intensity, Hue, Saturation） transform to obtain the intensity component of TM image. Then the SAR image and intensity component are decomposed using curvelet transform, and the curvelet coefficients are fused with different fusion rule in the different scales. Finally the fused image is obtained with inverse curvelet transform and inverse IHS transform. Some statistical parameters such as standard deviation, information entropy, correlation coefficient are used to evaluate the result. The visual result and statistical parameters show that this algorithm can not only preserve the spectral information well, but also increase the spatial detail information of the image, and its performance is better than that of the conventional algorithms, such as wavelet transform and IHS transform.