Research Of Panoramic Video Technology Based On Multiple View

Abstract: Abstract: the panoramic technique has become a hot topic in the field ofcomputer vision, video conferencing, virtual reality, video surveillance, tourism andother fields of scientific research, commercial applications with a wide range ofprospects.Panoramic video currently adopted by are generally equipped with panoramiclenses of professional camera equipment, such equipment is expensive, mainly usedin scientific research institutions or enterprises. To address this situation, this articlethrough the panoramic view of the existing technical in-depth studies designed apanoramic video acquisition method based on camera and experimentallydemonstrated the practicality of the system. This article first has categorized andsummary to existing image mosaicing, analysis the advantages and disadvantages ofa variety of algorithms, and selected suitable algorithm for this article. Secondly,analyzed the relationship between video and image, and stitching based on imagemosaicing applied to the video, at the same time resolved the synchronization ofvideo streams from software aspects. Finally, tested and analysis the performance ofthe system. The focus of this article read as follows:(1) Through in-depth research on panoramic technique, to the key technologiesinvolved a classification and summary was made. Design panorama video acquisitionmethod based on General camera, the system uses a modular design, and can meetthe individual needs of different users.(2) for video streams synchronization issues, in the case of hardware limited, solvefrom software, at the video processing stages created a buffer for each video frameand began to handle after the buffer has a certain number of video frames, which to acertain extent, ease the video stream unsynchronized phenomenon.(3) Registration in the video frame, used SIFT which has high robustness to rotatechange, light, and other features to operator for registration. And improved accordingthe necessary, experiments show that after be improved the algorithm havesignificantly improved in performance. In the video frame integration, using fusion method based on Gaussian function properties, which can effectively reduce the tonaldifference of fusion zone or ghost phenomena.(4) Finally, testing and analysis the whole system performance, and has beenextended as needed, on-board experiments showed that the feasibility and practicalityof the system.