Real-time Dynamic Splicing Display System Based On Infrared And Visible Images Fusion

In order to monitor the internal conditions of "EAST" device in real time,we propose a dynamic display system based on infrared images and visible images fusion by using multiple infrared cameras and visible cameras.Image fusion can eliminate redundant information in the images obtained in different bands,preserve complementary information as much as possible,and obtain more-rich and more comprehensive fusion images.Image stitching can break the limitations of cameras and use multiple sets of images with overlapping areas to splice into a large field of view image.At the same time,in order to synchronously reflect the internal operating conditions of the device,the system must meet real-time requirements.This paper analyzes some classic algorithms of image fusion and image stitching,evaluates and compares them.Combining specific application scenarios,the fusion of infrared images and visible images and the deficiencies in image stitching algorithms have been improved.At the same time,a complete set of real-time dynamic fusion and stitching display system is proposed and verified through experiments.1.Aiming at the problem of large time consumption in the image fusion process,this paper proposes a discrete wavelet fusion algorithm based on pixel threshold clipping.Only the specific area of visible image and infrared image is fused,and then the fused image is returned to the visible image,the amount of calculation is reduced and the time efficiency is improved.2.In the stitching process,the scenes are located in different planes.There is a difference in depth of field,resulting in large parallax.An image registration method based on multi-level transformation is proposed,and similarity transformation is performed on non-coinciding regions of the image to protect the image content.Can get a better stitching effect.3.For real-time requirements of the system,this paper performed off-line processing.Image transformation process parameters obtained in advance and call the parameters directly in the stitching process.In addition,using GPU-accelerated methods to meet real-time performance.