Design And Implementation Of Micro Infrared And Visible Fusion Visualization System

With the continuous development of technology and the increasing demand of society,image video has become the main source of information for people in nowadays society.In order to ob-tain comprehensive and complete scene information,multi-image sensor fusion system is widely used.Among them,infrared and visible image fusion system,as its key component,has gradually become a research topic for scholars.It is used for data integration of dual-source images,dis-carding redundant information,and retaining effective information.It currently plays an important role in military,security,and medical fields.This paper mainly designs and implements the mi-cro infrared and visible fusion visualization system,and focuses on the dual source image fusion algorithm based on multi-scale analysis and color transfer,which is improved and optimized the existing algorithms and achieves good fusion results in the system experiment.Firstly,the overall design of the micro infrared and visible fusion visualization system is com-pleted.According to the top-down design principle,the overall scheme design and system module division are carried out,and the software and hardware subsystems are designed respectively.The hardware subsystem takes ARM+DSP processor as the core to design the system structure,cir-cuit and so on.The software subsystem elaborates the system function,multi-threading and buffer,dual-core communication and so on.Secondly,the infrared and visible image fusion preprocessing is performed,the characteris-tics of infrared and visible images are analyzed.According to the data processing flow,distortion correction,filtering and denoising,image enhancement and image registration are sequentially per-formed.in the distortion correction,a calibration plate for simultaneously calibrating the dual source camera is designed for camera calibration.In image registration,edge features with high robustness and easy detection in dual source images are selected,which are unique to this system.Then,the dual source image fusion algorithm is studied,which is divided into gray and color image fusion.Gray-level image fusion is based on multi-scale analysis theory.According to the ba-sic framework of multi-scale fusion,non-downsampling Contourlet transform is selected as decom-position and reconstruction method,low-frequency subband saliency weighting and high-frequency subband adaptive pulse coupled neural network are used as fusion strategies.Color image fusion is based on color transfer theory,in a certain color space,the color statistical features of the reference image are transmitted to the target image to obtain a color image.At the same time,the target foreground area obtained from the infrared image is used as a mask to enhance the color contrast of the target.In order to ensure the real-time performance of the system,the two algorithms are optimized from both theoretical and program implementation.Finally,the system test and experimental analysis are carried out.The system test is used to verify whether the software and hardware of the design system are working properly.The exper-imental analysis is used to verify the effectiven.ess of the p.roposed image fusion algorithm.After introducing the image fusion quality evaluation method,the dual-source image fusion algorithm proposed in this paper is compared with the traditional algorithm of the same type on open data sets and system real capture data sets.The results show that the two image fusion algorithms proposed in this paper have prominent objectives,better details retention and practicability.