An Image Fusion Method Based on Multi-Resolution Segmentation

Image fusion method is a technology for synthesizing the information coming from multi original images of the same scene.Present region-based fusion methods can be classified into two classes: one class is to segment low-low frequency part of the highest layer to instruct the fusion of all layers;the other is multi-resolution segmentation method which is too difficult to meet the real-time requirement.Therefore,this paper develops a multi-resolution segmentation based fusion method.Its main characteristic is multi resolution segmentation.Its steps are: firstly it decomposes the original images by multi-resolution based on wavelet transform;secondly it segments low-low frequency part of each layer into different regions;finally it uses the region information of each layer to instruct the fusion of each layer,respectively.Simulation experiment indicates that the performance of this multi-resolution segmentation based image fusion method is better than those of the ordinary window-based fusion method and of the region-based fusion method only for segmentation of low-low frequency part of the highest layer.