Arm-Linux Based Perimeter Alert Algorithm Implementation Transplantation And Optimiza-Tion

With the acceleration of China’s urbanization process, public securi-ty insurance become more and more important. Video surveillance is an important means of ensuring public safety. However, traditional PC video monitoring system depend on manpower, bringing about a series of prob-lems like poor mobility, bad real-time performance, inaccuracy. With the development of Arm technology, calculation ability of embedded equipment increases significantly. Embedded network IP monitoring equipment combined with intelligent video surveillance become inevita-ble. This kind of equipment can break through geographical limita-tions of traditional PC monitoring, avoid human negligence, reduce secu-rity cost, and improve the quality of security. Therefore, embedded intel-ligent video IP monitoring equipment has a broad application prospect. This paper works on Arm-Linux based embedded intelligent perimeter alert system. The main research work embodies in the following aspects:1. In this paper, we study and analyze the connection between Arm-Linux embedded system and related equipment, determine the upstream-downstream interface and concrete requirements of Perim-eter Alert system. Based on the requirements, we design a complete Perimeter Alert system on embedded platform.2. In this paper, we study and analyze the existing Perimeter Alert sys-tem on PC platform and transplant it to Arm-Linux platform. Ac-cording to the demand for real-time performance and accuracy, this paper improves the performance of core algorithm like kalman filter and target tracking, reduces the computational complexity and im-proves tracking accuracy in worst cases.3. In this paper, we study the specific characteristics of the Arm-Cortex series kernel. According to the features, the Perimeter Alert algo-rithm is thoroughly optimized from algorithm level, code level and memory level and meets the need for real-time monitoring under common scene in the last place and reduce the processing time per frame to80ms in worst cases at the same time.4. In this paper, we design a demo platform which uses analog signal as both input and output based on the media process platform on Hi3516board by Hisilicon. It has met the needs of daily research and demonstration.