[Objectives] Plateau Pika (Ochotona curzoniae) is an endemic and key species in the alpine meadow ecosystem, and its behaviors of digging, foraging, and moving directly affect the health and stability of the alpine meadow. However, it is difficult to achieve continuous behavioral observation for several days to obtain all the data required for the biological rhythm observation of Plateau Pika. Due to the shortcoming of the traditional method,a Plateau Pika monitoring system is constructed by integrating image automatic acquisition, optimizing digital image and processing algorithm, and intelligent recognition technology, which realize non-contact, continuous, and digital monitoring of Plateau Pika. [Methods] According to the need for Plateau Pika activity rhythm monitoring, the experimental layout was designed with an area of 25 m × 30 m in Henan Mongolian Autonomous County, Qinghai Province. The Plateau Pika monitoring system is designed, and the optimized and improved frame difference method is proposed. [Results] Through the monitoring of the experimental area, the video image data of the recorded Plateau Pika is obtained. The Plateau Pika monitoring system meets the basic application requirements of observing Plateau Pika activities in the natural environment. The improved frame difference method is used to realize the detection of moving targets based on the complex image. The average accuracy rate, recall rate, and weighted average were 87.20%, 83.37%, and 0.85 (Table 2). The peak of ground activity of Plateau Pika appeared at 10:00 and 16:00, and it was inactive at night (Fig. 6), so it belonged to diurnal animals. [Conclusion] These results can meet the target detection requirements of Plateau Pika’s monitoring system, recording the activity rhythm and frequency, which provides new research ideas and effective technical means for Plateau Pika behavior research.