Font Size: a A A

Research On Infrared And Visible Image Fusion Algorithm

Posted on:2022-11-02Degree:MasterType:Thesis
Country:ChinaCandidate:Y J LiuFull Text:PDF
GTID:2518306749461244Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
Single source image sensor can only collect the single band information of the image.Multisource image sensor is mostly used to collect the image.Infrared and visible image fusion technology can improve the image resolution by using multi-source image sensors.However,there will be problems of noise amplification and detail loss.Therefore,this paper studies the infrared and visible image fusion technology from the following three aspects.Firstly,a dynamic range infrared image compression and detail enhancement algorithm based on contrast limited histogram equalization is proposed to solve the problems of inconsistent bit order between infrared and visible images and poor infrared image quality.The infrared image is divided into low spatial frequency layer and high spatial frequency layer by guided filtering;The limited contrast histogram equalization algorithm is used for image compression in low spatial frequency layer;For high spatial frequency layer,sharpening filter is used to enhance details;The processed low spatial frequency layer and high spatial frequency layer are linearly superimposed to obtain an 8-bit infrared detail enhanced image.Secondly,aiming at the poor adaptability of gray image to complex and changeable environment,a pseudo color fusion algorithm based on color transfer is proposed.Infrared and visible images are mapped directly to RGB channel to form a preliminary fusion image.Choose rainforest,desert,snow,ocean,architecture five different environment reference pictures for color transfer.The fused image and reference image are converted to LAB space,and the mean and standard deviation between them are calculated for color transfer.The pseudo color fused image is obtained by inverting back to RGB space.Finally,This paper uses FPGA to build a real-time fusion system of infrared and visible images.Infrared and infrared images can be input into FPGA through two-way Cameralink video interface.The FPGA logic resources such as look-up table,block ram are used to design video interface processing module,timing processing module,infrared image dynamic range compression and detail enhancement module,pseudo color fusion module,and finally the fusion image is output through HDMI video interface,Real time fusion of infrared and visible images is realized.
Keywords/Search Tags:Image fusion, High dynamic range compression, Color transfer, FPGA
PDF Full Text Request
Related items