Research On Key Technologies Of Panoramic Video Stitching

Panoramic image stitching is one of the fastest-growing branches of digital image processing.After years of development,related image stitching algorithms already have a mature theoretical framework system and are widely used in all aspects of our lives,even smartphones can take panoramic images easily.However,in the field of video stitching,due to a large amount of video data,the multi-video stream stitching operation is more complicated,and coupled with the prospect of moving in the video,the traditional image stitching technology alone cannot achieve good results.And it is difficult to meet the requirements of high quality and real-time,which limits the application of panoramic video stitching technology in various fields.The panoramic video stitching system constructed in this paper can stitch multiple video streams frame by frame through the overlapping area of adjacent videos to form a panoramic video with higher resolution and a wider field of view and output it in real-time.Panoramic video stitching technology can be widely used in all aspects of virtual reality,security monitoring,telemedicine,and social life.This article mainly studies image fusion algorithms and image brightness correction and their implementation on the embedded GPU platform.The main research work includes the following aspects:1.Researched and summarized the main process of panoramic image stitching,introduce the main principles and implementation methods of each step in detail,and choose for its application in video stitching,including image projection transformation,image color migration,and image Registration algorithm and image fusion algorithm,etc.2.Study an improved optimal suture fusion algorithm.Aiming at the problem that the traditional fusion algorithm can not effectively deal with ghosting in image stitching,this paper improves on the traditional optimal stitch fusion algorithm.When constructing the energy function,this method not only considers the difference in pixel intensity and gradient in the energy function of the traditional optimal stitching fusion algorithm,but also considers the difference in pixels on both sides of the stitched image after stitching,which makes the transition of the stitched image more natural after stitching,reduce the break of the texture and the appearance of ghosts.3.Study a brightness correction algorithm.Aiming at the problem that the brightness difference of the image to be stitched may make the panorama appear obvious light and dark transition traces in the fusion area,this paper studies a brightness correction algorithm based on histogram matching and feature point correction.This method combines global and local information,uses the overlapping area and feature point brightness information to perform brightness correction on the target image,reduces the brightness difference between the target image and the source image,and makes the stitched panorama have a more consistent visual effect on the image brightness.4.Established a panoramic video real-time stitching system.The real-time panoramic video stitching system constructed in this paper can process four channels of full HD video in real-time.In this paper,multi-threading technology is used to accelerate related processes,and GPU optimization of image mapping and image fusion is carried out to increase the processing speed and finally meet the requirements of real-time performance to ensure the real-time operation of the system.