Research On Difference Features-Driven Method Of Infrared Image For Mimic Fusion

The infrared intensity image relies on thermal radiation imaging to reflect the difference between the brightness of the target and the background,as well as the shape and position of the target.The infrared polarization image imaged by the polarization characteristic mainly reflects the edge details of the target and can also be detected.The hidden target of infrared stealth technology.Image fusion technology can be used to synthesize image information captured in different modalities or at different times in the same scene,thereby more comprehensively describing scene information and improving image visual effects.It is very important in target recognition,security detection,video surveillance,etc.application.When the shooting scene is different,the characteristics of the image are different.The existing infrared light intensity and polarization image fusion algorithm cannot dynamically adjust according to the image difference feature change,and the fusion needs cannot be analyzed according to the image,so that unnecessary fusion is performed.,Reduce efficiency.How to better distinguish the feature differences between different modal images,drive the fusion according to the difference of the images to be fused,and choose the appropriate fusion algorithm is of great significance and value for the development of adaptive fusion.According to the mimicry bionics thought,this paper proposes the infrared image difference feature driving method for mimic fusion based on the mimicry driving process of mimetic octopus.By analyzing the different properties of multiple features between bimodal infrared images,using the mimetic driving method,judging whether the images need to be fused according to the degree of difference,uniformity and complexity of the features,and the exact fusion requirements,provide guiding factors for image fusion.Therefore,as the image changes,the difference of the image features changes,and the image fusion method also changes to adapt to different fusion requirements.The main research contents of this paper are as follows:(1)Analysis of multi-simulated driving process and differential feature-driven fusion process: By analyzing the driving process of mimetic octopus,the driving process of the mimicry octopus is analyzed and modeled.The relationship between image difference and feature,the relationship between feature and fusion algorithm are analyzed,and the correspondence between the mimetic driving process and the difference feature driving method is established.(2)Multi-attribute characterization of feature differences in bimodal infrared images: image features are divided into three categories: brightness,texture and edge.From the difference of features,uniformity and complexity,feature differences and image differences are analyzed.The relationship provides data support for the establishment of subsequent driver methods.(3)Infrared image difference feature driving method for mimetic fusion: Based on the driving process of mimetic octopus,the differential feature driving method is divided into four stages: data acquisition,demand perception analysis,drive mode determination and fusion execution.Analyze the relationship between image differences and fusion requirements and classify the fusion needs.Each process is described in detail,and the data is extracted according to the driving mechanism,the fusion needs analysis and the feature-driven image fusion are used to verify the validity of the model through fusion experiments.