The continuous improvement in diversification, new-type orientation and low cost of navigation control system, the accurate measurement of the spinning aircraft flight attitude parameters has becomes a more and more urgent task. In view of the above problems, a novel attitude estimator for the spinning aircraft is proposed by using earth infrared radiation field. The attitude estimation system possesses several key advantages over the current designs in low cost, no need of moving parts, and being free from reliance on GPS or other state feedback. Firstly, the mechanism of earth infrared radiation field is described in detail, and an 8-14 m atmospheric window is selected as the study object. The land surface infrared radiation is calculated by the land surface temperature and emissivity. The sky infrared radiation is calculated through layered atmosphere by combing with the sky emissivity and infrared atmospheric transmittance. According to the calculations of land surface infrared radiation and sky infrared radiation, the mathematical model of earth infrared radiation field is established by combining with propagation law of infrared radiation in the atmosphere. Then the measurement model of thermopile sensors is derived, after analyzing the motion feature of spinning aircraft during the flight. The thermopile sensors convert the observed infrared radiation into an electrical signal well suited for onboard data acquisition. To explore the inner link between the thermopile sensor output and the spinning aircraft attitude information, the characteristics of the sensor output under different attitude angles and fields of view are studied. When the thermopile sensor characteristics are included, the fully developed model can be used to generate accurate sensor output as a function of attitude angle. Finally, the installation of the thermopile sensors on the spinning aircraft is designed, and the measurement model of onboard thermopile sensor is established. In order to improve the accuracy of attitude measurement, an extended Kalman filter is developed, which enables the estimating of real-time attitude angles and roll rate by using solely three-axis thermopile sensors as feedback. The result indicates that by using this high accurate algorithm, the pitch angle estimation error is within 0.02, the roll angle estimation error is within 0.1 and the roll rate estimation error is within 1 rad/s. The detection system is simple and practical, works stably, and can meet the requirements for spinning projectile attitude measurement. The attitude estimation system will provide a new method and theory for further developing the spinning aircraft state detection.