Font Size: a A A

Design And Implementation Of Video Correction And Transmission System For Fisheye Camera

Posted on:2020-08-03Degree:MasterType:Thesis
Country:ChinaCandidate:X CuiFull Text:PDF
GTID:2518306131464874Subject:Electronics and Communications Engineering
Abstract/Summary:PDF Full Text Request
The Fisheye lens,which can get more spatial information from a larger perspective because of its short focal length and large field of view,has been widely used in video surveillance,which needs omnidirectional visual information.However,the video taken with the fisheye lens will be seriously distorted because of its non-linear imaging model,which needs to be corrected.Accelerating the processing speed of fisheye video correction plays a vital role because it is difficult to achieve real-time performance on embedded devices for a large amount of computation.A fisheye correction algorithm based on GPU parallel computing technology has been designed and implemented in this paper,which can achieve real-time speed while ensuring computational performance.Firstly,a certain number of frames are extracted from the video stream.Calibration parameters are obtained through the camera calibration process.Then,the ideal and actual pixel values of the calibration points,as well as coordinate mapping tables between them,are obtained simultaneously through the calibrated coordinate map according to the imaging characteristics of the fisheye image.Finally,the coordinates of the pixels in each frame are calculated by using the coordinate mapping table to complete the video correction process,and the distortionless video stream is generated.The real-time processing of video correction is implemented in the case that the calibration parameters are acquired only once,and the coordinate mapping is computed by GPU parallel processing.Besides,an embedded video correction and transmission system based on the fisheye lens is implemented in this paper.The final system uses Nvidia Jetson TK1 as its hardware development platform while carrying a Linux system.The V4L2 driver,the correction module,the H.264 codec plug-in application,and Ethernet transmission are designed based on the GStreamer framework.The server and host clients based on Nvidia Jetson TK1 are built.The video stream is collected and corrected on the server-side and received by the client with VLC.The experiment results show that the final system can realize video stream acquisition,correction,and transmission in real-time performance.
Keywords/Search Tags:Fisheye video correction, GPU, Coordinate mapping, Transmission system
PDF Full Text Request
Related items