Research On High Resolution Remote Sense Image Fusion Method Based On Pixel Level

Remote sensing data fusion is an evolving technology that fuses multiple heterogeneous sensor data from the same region to obtain the enhanced information needed for decision-making.The application of data fusion often involves multiple disciplines.For specific purposes,the auxiliary and ground data are combined to improve the performance of data fusion.The remote sensing data fusion technology has effectively improved this problem.The fusion image can combine the features of different data sources to provide support for the analysis,extraction and application of image information.This has also been increasingly valued by scholars.This paper summarizes the current remote sensing image fusion technology.Through the concept of hierarchical classification,namely pixel layer,feature layer and decision layer,and discuss their principles and several representative fusion methods,mainly discussing optical panchromatic and multi-spectral data fusion methods.Until now,pixel-level fusion methods have focused on optical data fusion.From the perspective of pixel-level image fusion technology,the Panchromatic and Multi-spectral imagery of domestic high-resolution satellites with sub-meter level is taken as the research object,and the following research is mainly carried out:(1)The concept of remote sensing technology and image fusion is described comprehensively,summarizing and analyzing the basic theory of the purpose,level and process of remote sensing image fusion,and introducing the basic flow of image preprocessing at present,focusing on the geometric correction method.(2)In the view of the present methods of band selection(OIF)defect,the experimental analysis is carried out on the band combination,The results show that using 20% near-infrared band and the green band combination,the BJ-02 satellite image fusion effect is better.(3)Five traditional pixel-level fusion methods are discussed.They are wavelet transform fusion,principal component analysis(PCA)transform fusion,high-pass filter(HPF)transform fusion,Brovey transform fusion,and Gram-Schmidt(GS)for fusion simulation experiments,and analysis of these features and applicable environment of each method.(4)Based on analyzing and summarizing these traditional traditional methods,the fusion algorithm based on HPF transform and PCA transform and the fusion method based on wavelet transform and HPF transform are programmed in the Matlab environment to complete the fusion algorithm experiment,and the selection of quantitative evaluation index is determined.The fusion image is evaluated from the aspects of spectral fidelity,information volume and clarity.The experimental results show that the fusion method based on HPF transform and PCA transform and the fusion method based on wavelet transform and HPF transform can solve the spectral fidelity,information volume and sharpness of the fusion image,and an image with more abundant information.