Research And Implementation Of Embedded Network Video Software

With the development of computing technology and image processing technology, embedded network video applications which are supported both by embedded system technology and network video applications technology have been widely applied. But because of the resource constraints of embedded system and high resource requirements of embedded network video applications, a series of optimization methods should be researched to ensure the processing performance of embedded network video.In order to improve the capability of real-time processing network video by embedded software and allocating efficiently the computing resources in embedded systems, this thesis designs and implements an embedded software module architecture based on priority scheduling. It redivides functional modules of embedded network video software and assigns a priority to each module based on their timing relationships and real-time requirements, which can maximize the usage of embedded systems'computing resources. To solve the scheduling delay brought by the semaphore used to control concurrent access to shared resources, a non-blocking buffer management model is proposed and implemented. The test results show that these optimization methods of embedded network video software can improve the quality of processing network video.For network video quality may vary as the network load changes, network vido applications need some real-time methods to measure video transmission quality from the outside. To meet the requirement of measuring the delay of video transmitted in network, this thesis proposes a measurement method based on RTP to measure video delay, which transfer time information in the form of out-band data between server and client independent of video data. This measurement method has been implemented based on modifying the open source software of VLC and Live555. The test results show that this measurement method has a certain degree of accuracy.