Intelligent Video Surveillance Technology Research And Application In Agricultural IOT

Development of agriculture IOT is one of the symbol of agricultural modernization. As a part of video surveillance as the Internet of things, the importance of intelligent video surveillance technology rising. In spite of all aspects of agricultural production has a video surveillance system put into use, but most are the traditional video surveillance, only for artificial monitoring, research on intelligent video surveillance technology still has great research and practical value.This paper is devoted to the study of intelligent video monitoring technology research and application in the field of agriculture, to detect abnormal events focus on in the process of agricultural production. The main contents include the following three aspects:1.Presents an anomaly intrusion detection algorithm based on Motion Template. First discussed the target detection and tracking in static background and dynamic background, target detection and tracking in static background introduces the background difference, inter-frame difference and optical flow method, target detection and tracking of dynamic background introduces active contour tracking, feature based tracking, based on region-based tracking and based on the tracking model; then introduces the detailed calculation steps of abnormal intrusion detection algorithm based on the motion templates, real-time and accuracy comparison of target detection and tracking algorithm of the static background of quality, and considering the algorithm in the agricultural scene anomaly into practical detection applications.2.Proposed a detection algorithm of plant diseases and insect pests based onmulti-feature fusion and SVM classifier. First introduced the framework of the algorithm; then according to the selection of image features of plant leaf image, and introduces the method to extract these features; introduces the multi-feature fusion method based on Feature Pack; detection rate of plant disease and insect pest finally comparison of four kinds of single feature classifier and two kinds of multi-feature fusion classifier to the real scene image data. Conclusion the accurate rate of plant diseases and insect pests is based on feature fusion and SVM classifier for damage detection algorithm is higher, and meet the real-time surveillance requirements.3.To develop a multimedia information processing platform. This paperdescribes the purpose and design principle of platform developmentfirstly;and then carry out a needs analysis according to the actual application platform; then describes in detail the system architecture of the platform, including the part structure of the interface;finally introduced the database table structure and system interface. The platform has been applied.