Reserch On Key Technology Of Embedded Visual Surveillance System Based On OpenCV

Recently, the technology of visual surveillance develops rapidly. The equipment of surveillance with pre-processing in front-end based on embedded system is becoming a popular research topic now.As each application of traditional embedded monitor is designed and developed for a specific scene, there are some disadvantages in software development, such as long cycle, high costs and heavy workload. OpenCV is a good open-source computer vision library. It includes many image processing algorithms, which can be called directly in the process of development without repeating the programming, so that it can effectively improve the working efficiency. Therefore, the thesis mainly focuses on importing OpenCV to the embedded platform. It will be a good solution to the inadequate development problem of traditional embedded surveillance software, the main contents include:(1) This thesis analyzes the status and future trend of visual surveillance system and OpenCV, introduces the basic principles of embedded systems and features of OpenCV in detail, and proposes the selecting and designing plan for hardware and software in development platform of embedded system.(2) Analyzing the composition and system architecture of the OpenCV library functions in detail;studying the area and status of OpenCV in depth.In this paper,we focused on solving the crucial problems about transplanting OpenCV based on ARM platform,such as data type, memory alignment and byte order.At the same time,appropriate solutions were provided.(3) By combining both features of OpenCV library functions and embedded development, this thesis presents an idea of minimal cutting set, which can distinguish kernel module and redundant module of library function from the perspective of custom applications.(4) This thesis builds a visual surveillance development platform based on ARM and uses OpenCV to set up the programming environment. The Thesis also analyses the process of image capture for video4linux, designing a basic module for video monitor, and uses OpenCV function call camera to achieve a visual surveillance.(5) A series of test include integration, development of application and process of monitoring is implemented. The results can basicly attain the aim that transplanting OpenCV to embedded system for visual surveillance and development.