Remote sensing image fusion based on orientation information in nonsubsampled contourlet transform domain

摘要

For improving the traditional fusion algorithm quality and detail information, a fusion algorithm based on orientation information and pulse coupled neural networks (PCNN) in nonsubsampled contourlet transform (NSCT) domain has been proposed. Firstly, convert the multispectral (MS) image into intensity hue saturation (IHS) colour space. Match the histogram of panchromatic (PAN) image to the histogram of I component of MS image. Then decompose the I component and matched PAN image by NSCT, and apply orientation information combined PCNN fusion rules to NSCT coefficients. Reconstruct the fused I component by inverse NSCT transform. Finally, convert the fused MS image back to RGB space. A large number of experiment results have been done to prove that the method proposed in this paper gives better results than the other techniques used.