Multi-spectral images are of narrow-banded and high spectral resolution, however the transmission energy is low,
which result in a large scanning IFOV and a loss of spatial resolution; while SAR (Synthetic Aperture Radar) images, as
the angle reflection of buildings and low backscattering of waters, have excellent performance on texture structure and
water bodies. Therefore, adopting an appropriate fusion algorithm could obtain more accurate and abundant
information than any a single data.
Wavelet and IHS transform are complementary. An improved algorithm based on them applied to multi-source image
fusion is presented, which could overcome the disadvantages that classical algorithms have such as inconspicuous
improvement of space resolution, low level of information integration and serious spectral distortion. Intensity
component is first extracted from the multi-spectral image by IHS transform, and then I component and SAR image are
decomposed respectively by selected wavelet filter and decomposition layer. The modulation factor is gained through
regional energy measurement in sub-windows for the new I-component. Finally the fused image could be acquired
through inverse IHS transform. With different wavelet filters and decomposition layers, parameters are eventually fitted
on Coiflets for four layers with subjective and objective indicator criteria. Through regional energy fusion, we could
divide the smooth areas and marginal areas of the image in frequency domain, which could make a significative feature
measurement in smaller ranges. Experimental results indicate that this model would achieve an excellent effect on the
maintenance of spectral information in multi-spectral images as well as texture and edge in SAR images.