| The research of this thesis is mainly written for a study of moving target traced under the dynamic background which is based on ARM platform,furthermore,some improvement and optimized about tracing algorithms is also applied in this thesis,within the field of video surveillance and embedded development,the recognition and the tracing technology of moving target is an significant event,meanwhile, this research is also mean a lot to target recognition.by the way it’s necessary to make camera dynamically cause the single camera is not afford to trace the target continuously and entirely.however, the moving camera will be vibrate the whole detection background,that’s influence will effect much about the detection and tracing for the dynamic target,under this condition,the analysis for dynamic background been pretty important right now.The ARM11 platform and the Linux operating system is applied to this research and related algorithms are simulated by the Open CV transformed from OpenCV libraries,by the way,the NFS network file system is applied as the debug tools.In this thesis,the algorithms of recognition and tracing for these moving target under the dynamic background had been analyzed with detail, the organized dynamic background and the random dynamic background.In this thesis,these three situations will be separately verified and analyzed through different algorithms,under the first condition,this thesis applied the combination background subtraction method to analysis these images selected from the background libraries; under the second situation,images will be joint before the background subtraction applied in this algorithms; under the three situation which is the most complicated one,the method of background compensation with SURF characteristic is applied to build the background model.An optimized CamShift tracing algorithms is presented in this thesis proposed to decrease the complexity of time,to satisfy the requirement of Real-time in embedded platform. |
PDF Full Text Request