Study On Cross-Scale Fusion Of Multisource Images Within Network Environment

In this thesis we analyzed and studied the existing popular image fusion algorithm, and improved the cross-scale fusion algorithm with background of spatial using. Then a pyramid transform fusion and a wavelet transform fusion methods based on improved fusion rules were proposed. Meanwhile a cross-scale fusion system in network environment was implemented. Research and innovation are as follows:(1) Introduced and analyzed some multi-scale image fusion methods. Mainly the thesis introduced fusion methods based on the pyramid decomposition and wavelet transform. The former took Laplacian pyramid and SCDPT as example, while the latter were wavelet and dual-tree complex wavelet algorithm. Then we studied the advantages and disadvantages deeply.(2) Proposed a fusion rule based on relevance degree and region inner energy, which was applied to Laplacian pyramid fusion and SCDPT fusion methods. The algorithm flow of new methods, as well as experiment steps, were both showed in the thesis. Finally, the effectiveness of improved fusion rule was verified by comparative experiments on spatial pictures and standard images.(3) Proposed an improved fusion method based on cross-scale wavelet transform algorithm. For the characteristic of multi-scale of wavelet transform, we improved the fusion rules in both high pass and low pass, and also applied to dual-tree complex wavelet algorithm. The experiments showed that this method was good at maintain the orientation of image information. (4) A cross-scale fusion system in network environment is implemented. This system realized the functions of reading images information from sensor network, choosing fusion operator and fusion rules in client, feedback fusion result to users as well as showing the objective evaluation of fusion images.The proposed cross-scale fusion algorithms based on improved fusion rules made up for the disadvantages of traditional methods in contrast radio and detail information missing. The cross-scale fusion system solved problems of the gap between sensor network and fusion simulation platform, so that, the research laid foundation for cross-scale fusion within spatial background.