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

In remote sensing technology, in order to detect earth surface terrain features and need good spatial resolution image and panchromatic image can be very good to complete the objective. At the same time, in order to distinguish the different cover on the ground, the multi spectral image is also essential. And now satellite remote sensing can provide high spatial resolution panchromatic image and low spatial resolution multispectral image, and the image fusion can be achieved high resolution multispectral image. This is the significance of image fusion.This paper is mainly on the research of multi spectral image and panchromatic image fusion as the core and carry out. To solve this problem, the following contents are mainly studied:(1) Introduce and analysis of the purpose and significance of the multi spectral image and panchromatic image fusion in remote sensing technology,and I also introduce now at home and abroad research status.(2) The concept and level of image registration and image fusion are introduced,which are essential for image fusion. It also focuses on the evaluation criteria of image fusion field, including subjective evaluation index and objective evaluation index, including information entropy, mean value, standard deviation, average gradient, and so on. At the same time, four kinds of common traditional algorithms are analyzed, which are Brovey transform method, IHS transform method, PCA transform method and Gauss filter method.Finally, the four traditional fusion algorithms are simulated, and the simulation results are evaluated according to the evaluation method,and the two algorithms are compared with the IHS transform method and the PCA transform method.(3) To improve the IHS transform method. This paper introduces the theoretical knowledge of wavelet transform and its application in the field of image fusion. In view of the translation variation of wavelet transform, the SIDWT transform is introduced.An image fusion algorithm based on region segmentation is proposed, and the image fusion rules are detailed. The experimental procedure is given, and the simulation analysis and comparison are carried out to verify the effectiveness of the algorithm.(4) For the PCA transform method was improved. Describes the principle of Contourlet transform, including Laplasse transform and directional filter is introduced.Based on this, puts forward the transformation of PCA-Contourlet combined with Multi-objective Particle Optimization Algorithm of image fusion method, for contourlet transform of high and low frequency fusion rules are described, for multi-objective particle swarm optimization process are introduced, and gives the algorithm flow chart. Finally,through simulation experiments and subjective evaluation to prove the effectiveness of the algorithm.