Research And Implementation Of Fusion System For Infrared And Visible Image Based On FPGA

With the increase in application scenarios of various night vision equipment and security equipment,the information obtained is becoming more and more important.So the requirements for imaging technology are also increasing.Although current infrared imaging and visible light imaging technologies are already mature and stable,the technical defects caused by their innate imaging characteristics make them gradually unable to meet the actual needs.How to integrate the advantages of these two imaging technologies has become a research hots in the industry.This thesis firstly studied and analyzed the principles of infrared and visible light imaging and image digital processing methods,and used FPGA(Field Programmable Gate Array)as a development tool to complete the system design for fusion processing of these two imaging technologies.According to configuration requirements,the system is divided into three main modules: data acquisition,data processing and image display.Under the design based on the light path,the infrared and visible light image characteristic data are collected by dual lighting lenses;the analog-to-digital conversion of image data,video decoding and time-division multiplexing transmission are realized through the TW2867 chip.System separated image data and extracted the YCb Cr color information on the FPGA development platform.The median filter algorithm was used to reduce the noise of the separated data.The data of the non-common field of view of the two images were cut off,and then system used the quadratic interpolation method according to the cropping ratio to enlarge the two images to complete the image registration processing.The bionic color fusion algorithm improved based on the MIT method was used to complete the image fusion processing.Finally,the VGA protocol was used to realize the video image transmission and display.The experimental results show that the system designed in this thesis can collect clear and visible original images,and can effectively register two images,and can fuse infrared and visible images in real time and output them for display.After testing,the design retains the advantages of the two original images,while the target features are more prominent,the scene environment is clearer,and it is more conform to the visual physiological characteristics of human.The image SSIM obtained after fusion reached 0.6302,and the mutual information is 0.9154,which satisfies the design requirements well.