To enhance the dynamic performance and stability of spacecraft attitude measurement, a commonly adopted approach is to integrate a star sensor with an inertial gyroscope. While an integrated sensor system offers benefits in high-precision and high-dynamic scenarios, it also introduces errors in the extrinsic parameters of the camera and gyroscope, directly contributing to increased fixed error in attitude determination. Hence, real-time correction of integrated system parameters becomes crucial for enhancing the accuracy of attitude estimation. However, a deep coupling relationship exists between the camera's principal point and the extrinsic parameters of the two sensors, unified calibration of system parameters increases the computational time to decouple all the parameters. In response to this challenge, a stepwise calibration method for system parameters is proposed. Firstly, the calibration of camera intrinsic parameters leverages the principle of invariant interior star angles, which are independent of extrinsic parameters. Secondly, based on the observation vector error model of the integrated sensor, observation equations and state equations are constructed, followed by proposing a 4-position maneuver strategy based on observability analysis to achieve rapid decoupling of gyroscope bias, attitude error, and extrinsic parameters. Simulation results affirm the effectiveness of the proposed method, extrinsic parameter errors reaching the arc-second level.