Robust Image Fusion Alogrithm Reasearch Based On Layered Information

With many application platforms preferring to obtain images with multi-sensor for high performance,image fusion technology which aims to merge multi-sensor images into single more comprehensive fused image is becoming more and more important.This technology has played a crucial role in multi-sensor equipment and been applied into photography,remote sense,military,medicine and etc.When the application scenarios become more complicated,traditional image fusion technology cannot meet the require-ment of project and application entirely due to lack of ability to handle interference.This paper focuses on robust fusion technology for noisy misaligned images in sin-gle channel scenarios,such as infrared-visible,multi-focus and medicine.This paper makes a derivation of theory and designs a novel general scheme of correlated-particular layer based robust image fusion.The theory and scheme improving robustness,adaptabil-ity and adjustability of image fusion simultaneously.Cross presentation frames,which improves accuracy and avoids mixture of information extract module,are proposed re-spectively in pixel domain and sparse domain.In layer decision module,minimum-ratio method and max-sparse-coefficient method divide image into correlated components and particular component effectively with considering of domain's characteristics.With in-tegrating the informative distribution of correlated and particular layers,the context of two images are merged into fused image successfully.Therefore,proposed algorithms achieve their robustness through utilizing the traits of noise and misalignment within the fusion scheme.Proposed two algorithms are held comparison with some excellent image fusion algorithm in scenario which doesn't contain any noise or misalignment.The objective evaluation proves my method is better than state-of-the-art methods.As for scenarios with noise and misalignment,the image fusion algorithms based on correlated-particular layer perform impressive robustness.In conclusion,this paper has attain the expected research goal.