Research On Remote Sensing Imagem Fusion Technology Of MS And PAN

With the development of remote sensing and communication technology, more and more multi-temporal, multi-spectral and multi-resolution remote sensing image data are provided by different kinds of remote sensors. How to more efficiently, reliably, accurately and synthetically make use of the iamge data become research focus. Image fusion technology can reach this target, so it has a wide range of applications and rapid development. The major researches of this thesis is the image fusion of multispectral and panchromatic images, systematically summarize the background, history and current situation of image fusion, and points out the difficulties in the algorithm and evaluation system reaearching and researching direction the next step.Due to fused image with high-resolution based on IHS transform and with good spectral feature based on wave transform, a image fusion algorithm based on IHS and wavelet transform has been proposed. Wavelet coefficients fusion rule, such as in some conventional algorithms, is not been key point in the algorithm. The focus is the introduction of panchromatic image and original multi-spectral image brightness component composition varying degrees through adjusting parameters in the last IHS inverse transform process, which reduce information loss comparing with algorithms based on wavelet coefficients fusion rule a certain extent, to gain a good balance between spatial resolution and maintaining spectral information and promotion.In addition, the algorithm has the advantages of a small calculation amount and simplicity. The New algorithm has been tested on various multispectral and panchromatic satellite remote sensing images on FORMOSAT-Ⅱ, IKONOS, QuickBird, SPOT, LANDSAT, and experimental results demonstrate that it is simple, right and effective. In addition, this article analyzed the image fusion technique using steepest descent method to carry out fusing to high spatial resolution panchromatic (PAN) images and low spatial resolution multi-spectral (MS) images of IKONOS. Aiming at the disadvantage that clarity of fusion image is sacrificed seriously, an improved algorithm is proposed on this basis. To replace constant parameters added as weight with R, G and B bands images in the energy equation of original algorithm, matrix parameters which are set up in accordance with the proportion of R, G, and B component value of each pixels of MS image (bands:R, G and B) are used in the improved algorithm. The advantages of improved algorithm are that hue of MS image is maintained fully, characteristics of spectrum are preserved well, and the clarity of fusion image is improved substantially at the same time. The results of simulation experiment demonstrated the effectiveness of the improved algorithm.