Design And Implementation Of Warehouse And Forest Supervision System Based On Moving Target Detection

Carrying out agricultural product pledge supervision business can effectively alleviate the financing difficulties of agricultural SMEs and rural economic organizations,and is one of the effective ways to solve the "three rural" issues.The core of the agricultural product pledge supervision business is the accurate detection of the moving targets in the warehouse and the fruit tress area.The traditional pledge monitoring method requires long-term manual labor for video surveillance,which has the problems of untimely supervision and low efficiency.In this thesis,video surveillance-based moving target detection technology is adopted to achieve accurate real-time monitoring of moving targets in warehouses and fruit trees.Among them,in the detection of moving targets in the fruit tress area,due to the complexity of the ground,there are geometric changes and occlusions of the target to be detected,resulting in the target being easily missed or incompletely detected,and the detection accuracy is low.Therefore,the warehouse and tress area supervision system based on mobile target detection is designed and implemented in the thesis,which has important application value for the development of agricultural product pledge supervision business.The main work of this thesis is as follows:(1)The orchard warehouse pledge supervision system is designed and implemented.Aiming at the problem of untimely supervision and low efficiency of the pledges in the warehouse,the pledges and moving human bodies in the warehouse are taken as research objects to study the movement of target objects and detection methods of human body movements.The edge detection and the inter-frame difference method are mainly used to extract the contour of the moving target,and then the real-time accurate detection of whether the pledge moves through feature matching is realized.Consider the potential regulatory risks that human movement may bring,and carry out human movement detection methods.(2)The moving target detection system in the fruit tress area is designed and implemented.For the problem of missing or missed detection caused by occlusion and attitude change of moving targets in fruit trees,the detection method of moving targets in fruit trees is studied.The inter-frame difference,adaptive threshold segmentation and connected domain label segmentation methods are used to extract the contour of the moving target,construct the moving target feature vector.Finally,the Naive Bayes algorithm is used to determine the difference between the two frames of the image before and after,to achieve accurate detection of moving targets in fruit tress areas.(3)The warehouse and tress area supervision system is designed and implemented.According to the supervision needs of Qin Orange Orchard Warehouse and Orchard District,the warehouse and tress district supervision system is designed and implemented in this thesis,including the Orchard Warehouse Pledge Supervision System and the Orchard District Movement Target Detection System,which achieves the accuracy of the warehouse and orchard district movement target Detection.In summary,the warehouse and tress area supervision system is designed and implemented in this thesis.The feasibility of the system is verified through experimental analysis of different scenarios.The experimental results show that the system detection accuracy and real-time performance designed and implemented in this thesis meet the actual needs of the project and have certain engineering application value.