Abstract: Video surveillance in field agriculture has been widely applied in monitoring animal and plant growth, illegal invasion, pests, and disease, among which moving target detection is a key issue in the field of Agricultural video surveillance. Based on the strong ambient light interference in the field, rapid change of speed of moving targets, and quick changes in scenes, it was analyzed the composition of the background template, tried to use the results of the frame differential method to adjust the proportion of background template, and proposed a new background subtraction method fusion frame differential method of moving object detection algorithm to enhance the accuracy and the rate of the algorithm.In addition, an intelligent agricultural video monitoring system based on an embedded operating platform was developed to solve the problems existing in the field of agricultural video monitoring, such as the uselessness of a large capacity of video content, inconvenient layout, lack of utility power.The system had embedded processor, camera, memory, mobile communication, solar power, and other modules. We used a new algorithm in this paper. Compared with the traditional fusion algorithm, the accuracy was 0.55% higher, the storage capacity of surveillance video was 33.3% lower, and the battery life was 26.4% higher.