Research On Color Video Generation Technology Of Multi-band Imaging System

In recent years,artificial intelligence and network communication technology have gradually developed,and the application scope of video surveillance and image processing technology has also expanded accordingly.Traditional monitoring equipment mostly uses single-band,fixed-position cameras,and the monitoring range is relatively small.When the weather conditions are poor,using infrared cameras can only generate grayscale images,which cannot meet the high-resolution and high-information video monitoring requirements.With the zoom-type high-resolution camera,a wide range of monitoring angles can be flexibly selected.The colorization of the infrared video background can increase the contrast between the infrared foreground moving target and the visible light background,and the image information is richer.Therefore,zoom-based multi-band imaging is proposed and implemented Color video construction system.Take multiple visible background images of the surveillance area during daytime with high visibility as the source image of the infrared to color background,respectively use the YOLO algorithm and HOG+SVM algorithm to detect moving obstacles such as vehicles and pedestrians in the image,and use the same position image replacement method Obstacle elimination is performed to reduce the interference to the moving target after fusion with the infrared moving target.The camera uses an infrared background image with any field of view and focal length in the infrared band shooting area under poor weather conditions.By adding the same ratio conversion function,the panoramic stitching and template matching algorithm are used together to extract the colored background image.Use the SIFT algorithm to panoramicly.stitch the barrier-free visible light background image into a panoramic image.The infrared background image and the panoramic image are transformed in the same proportion,and then use the template matching algorithm to automatically register with the panoramic image and extract the same color as the infrared background image.The background image realizes the conversion of infrared background to colorized background at different focal lengths.The color background image is used as the video background,and the infrared moving target is extracted from the infrared video frame based on the combination of three-frame difference and Gaussian modeling,and the mask fusion algorithm is used for fusion to realize the colorized video construction.The experimental results show that the video frames fused by the algorithm of this system can meet the requirements of all-weather color video construction under zoom multi-band camera.Use Visual Studio 2015 software development tools,combined with MFC(Microsoft Foundation Classes)as the application display framework,cooperate with the OpenCV computer vision library for digital image processing,develop a complete multi-band imaging color video generation system software platform,this platform is suitable for It is used for fusion imaging of various infrared band images and visible light background images.The thesis uses infrared LED cameras to perform functional tests under actual field environments in different locations to prove the availability of the system.The color video output by the system can ensure that the infrared surveillance video taken at any focal length and field of view can realize the background color processing,the processed background scene information is more abundant,and the infrared moving target is compared with the color background to locate the position and movement status More obviously,the monitoring personnel can watch the color background video in any wave band and any focal length in real time through the system.