Research On Timestamp Removal And Replantation For Video And Its Application In Education

Multi-camera video fusion is the most important subsystems of intelligent educational systems.The monitoring of large venues on campus,panorama video recording in the classroom,and educational robot multi-view visual perception system,they all need to capture the video synchronously and synthesize the panoramic video by multiple cameras.During this process,timestamp removing and replantation is a challenging problem.When each video has a superimposed timestamp which is not removed,the merged panorama video by multiple videos will appear multiple timestamps in a frame.These timestamps may affect visual effects and cover some interested objects.To produce a good quality of panorama video,the timestamp need to be removed from each individual video and replanted into a merged panorama video.The main research problem of this paper is how to efficiently remove the timestamp of each individual video and replant a timestamp into a merged panorama video.The research objective is to research into difficulty and problems of timestamp removal and replantation process for the monitoring of large venues on campus,and propose a series of methods and key technology to solve that,and apply these methods in the education system.Our research content includes six aspects,(1)timestamp localization;(2)timestamp region removal and restoration;(3)multi-camera color correction;(4)timestamp recognition;(5)panorama video generation;(6)panoramic video system based on educational robot are developed.Through the research of the above six contents,this paper has made some contributions to the academic,and the main contributions are as follows:For timestamp region removal and restoration problem,the traditional image inpainting methods can fix the image by using the pixel or texture in the nearest neighbor region instead of a real image.To solve this difficult problem,we make the portion covered by timestamps to appear in the adjacent frame by active control PTZ camera shift periodically,and then realize timestamp region removal and restoration using a fast image translation estimation algorithm and temporal-neighbor pixel recovery method.The fast image translation estimation algorithm uses PTZ motion estimation combined with global histogram-based image matching to calculate the translation between two frames.The algorithm has a lower computational complexity and better real-time performance,because the translation between two frames and overlapping area are estimated by the azimuth information of the two preset points.The temporal-neighbor pixel recovery method can locate the portion covered by timestamps in adjacent frames,according to the fast image translation estimation result,and then repair the timestamp region using image fusion technology.Experimental results show that this method can remove the timestamp with high scene consistency.For multi-camera video color correction problem,we develop a mixed color correction technology by combing the region matching with the gamma correction and the linear correction to solve the problem of color correction in irregular overlapping regions.The experimental results show that the color of the target image is highly consistent with the reference image by using this technique for irregular overlapping regions.In the educational robot platform,we design and implement a panoramic video system.The system can achieve the goal for timestamp removal and replantation.