Research On Fusion Algorithm Of Multi-spectral And Panchromatic Remote Sensing Images

With all kinds of new remote sensing satellite sensors are increasing, the availablemulti-spectral and high spatial resolution remote sensing image data are also increasinglyrising dramatically. How to extract and integrate effective information reasonably from theremote sensing image data , in order to overcome the limitations of a single image itself,and finally achieve the goal of the accurate expression of the observation target, becomesthe focus of the study. Remote sensing image fusion technique is an effective means tomeet this requirement, so it has a wide range of application and development. Moreover, itplays a key role in the further analysis and processing of the image and also in theidentification and testing of target object.The pixel-level image fusion algorithm which can improve the spatial characteristicsof the multi-spectral image as much as possible with less spectral distortion as theprerequisite is explained in this paper, multi-spectral image and panchromatic image takenas an example. The article mainly includes the following three aspects:Firstly, based on the research background, significance and the present situation,remote sensing image fusion technology is discussed. The existing problem in fusionalgorithm and the future direction of the research are also pointed out. Furthermore,evaluation criteria of image fusion and several kinds of traditional fusion methods arepresented. Through the simulation experiment, qualitative and quantitative analysis aboutthe advantages and disadvantages of these traditional methods and application occasionare given.Secondly, multiresolution characteristic and advantages of wavelet is analysed.Aiming at shortage ofàtrous wavelet transform, a image fusion based onàtrous wavelettransform and MOPSO is put forward. The experimental results show that the fusionalgorithm enhances the spatial geometric features and maintains the spectral radiativeproperties excellently.Thirdly, a remote sensing image fusion scheme is proposed, based on non- subsampled directional filter bank included in NSCT which provides a full shift-invariantdirectional multi-resolution trait, combined with the superiority ofàtrous wavelettransform. According to the high-frequency coefficients obtained by the transformation,this paper proposes a new method of minutia extraction. The result of the experimentvalidates the effectiveness of the proposed algorithm.