Design And Implementation Of Front-end Embedded Software Of Intelligent Video Surveillance System Based On SoC

Traditional video surveillance system completely depends on humans to realize"Monitoring"and"Control", which is called unintelligent video surveillance system or passive video surveillance system. It gives rise to many problems, such as attention and eyesight tiredness, network resources waste, storage resources waste and search difficulty. Thus, it can not meet the increasingly growing demand on security monitoring for public security, transportation and banks. With the rapid development in wireless access technology, SoC technology and intelligent video analysis technology, the intelligent video surveillance system, which has function in abnormal events perception, becomes a tendency for video surveillance system and hence called next generation video surveillance technology.This thesis, according to the trait that the key of intelligent video surveillance system lies in the intelligence of the front-end, designs and implements the front-end software of embedded intelligent video surveillance based on SoC. The front-end system, using as core platform the embedded processor i.MX27 of Freescale Company and embedded Linux operating system, not only has functions of video image acquisition, compression, storage and transmission as traditional video surveillance system requires, but adds new functions of abnormal events perception based on moving object detection and automatic alarm interaction, which implements of basic intelligent video surveillance.The thesis first elaborates on the current situation and development trend of video surveillance system, and then introduces the overall design of the selected intelligent video surveillance front-end system. The hardware of the front-end system takes embedded processor i.MX27 as the core device and forms the main board with peripheral hardware circuit; its operating system is embedded Linux. The thesis completes the transplantation of Bootloader, kernel customization, construction of root file system and design and implementation of the driver of video capture; then it accomplishes the design and development of application programming for front-end of intelligent video surveillance on embedded Linux operating system. According to different functions, it divides the applications into video processing subsystem and auxiliary control subsystem; each subsystem adopts modularity design, and uses various programming methods to ensure the system integrality and operating stability. In the end, the paper does research on the core algorithm of intelligent video surveillance applied in this thesis: Moving Object Detection Algorithm; based on advantages and disadvantages of several common algorithms, embarking from Frame Difference and Background Difference, and combining processing capacity of the hardware platform, we design and implement a low-complexity and relatively high-accuracy algorithm.The thesis designs and implements the front-end software of intelligent video surveillance and carries on system coordination and tests with user's terminal of surveillance center based on PC. The tests and experiments indicate the system has good capacity on anomaly perception and alarm for perimeter intrusion and objects removal; it reaches the aim of the design and preliminarily realizes the digitalization, networking, wireless and intellectualization of video surveillance system.