Remote sensing image fusion using intensity-hue-saturation transform and steerable pyramid transform

Remote sensing image fusion (also known as Pan-sharpening) can be used to get a high spatial resolution multispectral image by fusing a low spatial resolution multispectral image with a high-spatial resolution panchromatic image of the same scene. We developed a pan-sharpening algorithm by applying the IHS transform and SPT (steerable pyramid transform). SPT transformation is used to decompose the panchromatic image (histogram matched with the multispectral images) into SPT coefficients. The composite SPT coefficients are obtained by substituting the subband of the SPT coefficients in the lowest frequency domain with the multispectral image's IHS intensity proportion in the colour space. Inverse SPT on the composite SPT coefficients is performed to produce a new intensity component. Finally, through the IHS inverse colour space transform, we can get the pan-sharpened multispectral image. Pan-sharpening results using real data show the pan-sharpening performance of the developed algorithm is better than the performance of the traditional methods, such as the algorithms based on IHS colour transform, DWT muliti-resolution wavelet transform or DWFT multi-resolution wavelet frame transform.