Research And Implementation On Intelligent Surveillance System Based On Embedded System

Recently, people care more about security. Whether government or enterprise invests more and more in the security field in order to ensure urban stability and security, however, the traditional surveillance system is faced with ineffective man’s monitoring video. Therefore, the research of intelligent surveillance system has become a hot spot.This thesis aims to propose a solution of an intelligent video surveillance system. The video surveillance system is composed with a DSP embedded system as intelligent analysis end (AE) and a PC as controlling end (CE), and AE communicates with CE through IP network about video and state and controlling information. It has the following functions:CE can send configuration order to AE. Meanwhile, AE can respond to CE about its working state. Furthermore, it can capture video and compress it in order to transmit it to CE. Finally, it can analyze the video content to see if a person exists, and then send an alert signal to CE.This thesis uses DM6437 embedded platform as AE which can capture and compress video and detect human body for its great ability of operation. DVSDK is short for Digital Video Software Development Kit, which is provided by Texas Instruments. AE uses CODEC ENGINE in DVSDK as encoder. Meanwhile, this thesis realizes human detection based on head-shoulder feature on this embedded system. The software on CE is developed by Visual C++6.0. It can decode H.264 stream by JM decoder and display video through SDL.Finally, this thesis proposes a surveillance system based on embedded system.