Algorithm For Remote Sensing Image Fusion Based On The NSCT

Image fusion technology is an important topic in the field of computer vision and imageunderstanding, widely used in areas of military, computer vision, remote sensing application.This article research on remote sensing image registration and fusion, main research contentsare as follows:1. According to the analysis of the image registration algorithm based on crosscorrelation, mutual information, wavelet transform, put forward the method based onnon-sampling contourlet transform and mutual information. Firstly, respectively decomposetwo images to three layers with NSCT, secondly, register approximate component from thehighest layer of image approximate component to the lowest layer, finally, get the new image.The experimental results verify the effectiveness and accuracy of the registration method.2. Comparative and analysis fusion method based on algebraic operations, based on colorspace transformation, based on multi-resolution analysis method.For multispectral andpanchromatic images, presented methods of image fusion based on NSCT and IHS.Firstly,register two image, secondly, convert multispectral image from RGB space to IHS space andget I component, thirdly, decompose panchromatic image and I component to three layerswith NSCT, and get the low frequency and high frequency images. For low frequency image,we adopt the fusion rule of regional energy adaptive weighting. For high frequency image, weadopt the fusion rule of local energy. These experiments proved this approach would be betterthan other methods.