In order to save storage space of video surveillance systems

and to reduce redundancy of video recordings content

an embedded wireless video surveillance system is designed based on the Qt framework. On the basis of S5 PV210 microprocessor hardware platform

camera control cloud platform is set up. Video images are sent to client-end under WiFi environment by RTP/RTCP protocol. The system Gaussian mixed model is used to extract foreground of target and combines dynamic inter-frame difference method to detect whether the monitoring region changes

and only record effect dynamic pictures in real time. Remote angle control of the camera cradle head is carried out by TCP protocol. The experimental results show that the video stream detection algorithm takes an average of 43 m

which meets the requirement of real-time performance and saves the storage space.