Font Size: a A A

Research On Indoor Video Monitoring Technology For Moving Object Based On Arm

Posted on:2013-05-13Degree:MasterType:Thesis
Country:ChinaCandidate:B M LiFull Text:PDF
GTID:2248330362962638Subject:Instrumentation engineering
Abstract/Summary:PDF Full Text Request
Video monitoring is an important part of security system. With the development ofembedded, multimedia and network technology, the video monitoring technology has alsomade great improvement. It is people’s first choice to construct an efficient, stable andflexible embedded video monitoring system which integrates the embedded, multimediaand network technology. This project takes the indoor video monitoring for moving objectas the applications background, and focus on the research of video monitoring technology.Based on embedded platform, the video monitoring algorithms take the ARMprocessor as the core and run in the embedded Linux platform. The moving object and itsmotion locus will be detected from the videos collected by two CMOS cameras throughseveral intelligent algorithms, such as the camera calibration, moving object detecting andmatching algorithm, and three-dimensional localization algorithm. Whenever the movingobject breaks in, the monitoring system can inform the user accurately and timely, andshow the abnormal picture with either the graphical interface or the network transmitmethod.First, on the ground of some common algorithms of motion detection, an improvedalgorithm was provided for the motion detection, which combines the backgroundsubtraction with frame difference. An improved feature matching algorithm was alsopresented after having compared the common feather matching algorithms, which includestwo parts: One is to use SURF match points detection and Harris corner detection toextract feature points in the rigion of interest, the other is to use the improved Ransacalgorithm for wiping out the wrong-matching.Secondly, binocular stereo vision cameras were calibrated with a calibration panel.The 3-D reconstructing technology and intrinsic parameters were used to acquire the 3-Dlocation of the moving target center of scene.Thirdly, the embedded Linux development platform was constructed. The wholeprocess included not only the build of cross development environment, but also thetransplant of several codes, such as the U-Boot, Linux kernel, root file system, Qt graphicsdevelopment tools and so on.Finally, a Qt graphical user interface with functions of camera video capture and image display was designed. A file transfer module based on TCP protocol was made.Besides, the performances of all the algorithms related to the project were analyzed withsome application tests under embedded environment.
Keywords/Search Tags:video monitoring, motion detection, embedded Linux, camera calibration, SURF algorithm
PDF Full Text Request
Related items