This article is to study the system structure scheme based on Zigbee wireless transmission, and complete the overall design of the system scheme on this basis. Human motion capture systems are widely used in the creation of film and television works, motion analysis, video games, rehabilitation medicine and other fields. This article discusses the design and implementation of a human motion capture system based on MEMS sensors and Zigbee networks. The system can be installed on the human body Multiple sensor nodes in various parts obtain the movement information of the human body, and use sensor network technology to aggregate these data and upload them to the host computer. First, this article introduces the characteristics of angular velocity sensors, acceleration sensors, magneto resistive sensors and Zigbee networks. Then, this article explains the overall structure of the system, and from a theoretical point of view, explains how the system uses angular velocity sensors, acceleration sensors, magneto resistive sensors and Zigbee networks to achieve human motion capture. This part focuses on including vector observation methods and angular velocity Two posture capture methods including the integration method, and their advantages and disadvantages are analyzed. To achieve the complementary advantages of the two algorithms, a data fusion algorithm based on complementary filtering is introduced and optimized appropriately. In addition, this article also introduces the networking principles and optimization schemes of the Zigbee network in this section. After this, this article explains in detail the system hardware structure, chip selection scheme, circuit design scheme,software workflow and implementation of core programs Method. Finally, this article shows the effect of the actual work of the system, and compares it with the theory to verify the feasibility of the theory. Based on the research of MEMS sensor measurement unit and algorithm, a Zigbee-based wireless transmission test system was established. LabVIEW software with functions of data reception, attitude angle calculation, trajectory calculation, eigenvalue extraction, BP neural network recognition, display and data saving was designed and tested the whole system functions. The test results show that the wireless data transmission of Zigbee network is normal, the data detection and processing programs of the host computer are stable, and the correct identification of the human body’s motion state can be realized. The results show that compared with the existing research, our research has increased its efficiency by 10%, and its accuracy has increased by nearly 15%.