Font Size: a A A

Research On Key Technology Of Fast Stitching For Unmanned Aerial Vehicle Video Frame

Posted on:2017-03-16Degree:MasterType:Thesis
Country:ChinaCandidate:D L YangFull Text:PDF
GTID:2348330488462543Subject:Mathematics
Abstract/Summary:PDF Full Text Request
Taking advantage of unmanned aerial vehicle video to make large scale static image has many usages in military reconnaissance,disaster relief and other applications by master global information,it is a key technology of unmanned aerial vehicle application.At present,stitching for adjacent frame of video is an obtain way to get large scale image of big scene,therefore,we should research the rapid processing technology of the UAV video according to the characteris-tics of UAV video frames.Adjacent frame stitching is combined of frame matching and frame stitching.Rapid frame matching is the key technology to UAV video fast processing and it is also the presupposition of frame stitching.After frame matching,stitching from frame to frame will cost much time and lead to obvious deformation because of the huge data,so in order to solve the problem,we research these works such as rapid matching between adjacent,key frames' se-lection,the judgment of the routes and frame stitching.The main work of this paper is as fol-lows:1?In view of the fast matching of adjacent frame,we give the rapid matching algorithm based on PID prediction by take advantage of adjacent frame's characteristics.Adjacent frames have a high similarity because of they can obtain within thirty milliseconds.Considering of the video imaging process,frame matching can be regarded as a big parallax gradually moved into image matching problem.Setting up PID model can efficiently improve the algorithm efficiency by predict the corresponding point's coordinates in the next frame based on corresponding points displacement in the first few frames.Therefore,researching PID prediction can reduce the size of search window,so we can have a higher efficiency.2?Face to the problem that video has a large amount of frames and the cumulative error after stitching many times,this paper give an algorithm by combine projective transformation and similarity transformation to stitch adjacent key frames.We tectonic energy equation for the dividing line between two kinds of transformation and finishing stitching between the key frames.Finally,we summarized the work of this paper and analyze the future study.
Keywords/Search Tags:PID Prediction, Key Frame, Projective Warp, Similarity Warp, Stitching Seam, Panoramic Image Stitching
PDF Full Text Request
Related items