| In recent years,with the rapid development of remote sensing image technology,multisource remote sensing image fusion has become one of research a hot focus in the field.Remote sensing image fusion has become one of the indispensable technique in the field of image processing,and in many fields such as agriculture,military,land planning plays an important application.However,in many practical applications,the requirements of spatial information and spectral characteristics of remote sensing images are not exactly the same.At present,the method of remote sensing image fusion is difficult to accurately and efficiently complete the sub regional image fusion according to the different needs of this region.To solve this problem,the paper is based on the reading and researching of related methods both at home and abroad,makes a deep research on remote sensing image fusion process 、pretreatment technology、the method of Image registration、the method of Image fusion.The main work is as follows:In remote sensing image automatic registration,there are some problems,for example,there are differences between texture image itself and unevenness in feature points extracting.To this problem,the paper studies high precision automatic registration method of multi-source remote sensing image.First of all,the Harris operator was improved to extract feature points covering a wider range;And then,the image is divided into blocks,for each block,using the improved Harris operator to extract feature points,and then use the block iteration to eliminate wrong feature points,ensuring the distribution of feature points is uniformly and non redundant in the whole image;Secondly the paper puts forward the pyramid-bidirectional-least squares matching algorithm for image matching,completed selection of the final corresponding points;Finally the paper makes use of the TIN to complete accurately correction.In Image fusion,on the study of the significance analysis model and the method of image fusion.the paper proposed a new salient analysis model is called IR(adaptive-radiussearch)salient analysis model,which realized the identification and classification of the salient regions and non salient regions.and for these two areas,combined with HIS and wavelet transform fusion algorithm,the paper puts forward an improved adaptive HIS fusion method based on average gradient weighting,which achieved the fusion of the road,farmland,residential area and other salient regions,and the method has better keeping the spatial details information of its abundant space;Meanwhile,the fusion method of HIS+WT transform is proposed,which realized the fusion of mountain,forest land,and other areas are not salient,retaining more spectral information.The method completed the adaptive remote sensing image fusion of different regions.The experimental results show that,the registration points obtained by this method are uniform and reasonable,and the registration accuracy is high;The adaptive fusion algorithm can make the fused remote sensing images keep high spatial detail information and rich spectral information.The proposed method of the paper provides certain theoretical basis and application value for the research on agricultural science,forest planning and future forest remote sensing image classification,identification and so on. |
