Magnetometer is the main instrument to determine the attitude of micro satellite. Magnetometer accuracy is an important indicator which influences the micro satellite attitude performance. In order to improve magnet estimation accuracy, in-orbit calibration based on multi-parameter is utilized in two stages to estimate the error of Magnetometer`s bias and scale factor. The results can be used to compensate the attitude algorithm, which improves the navigation accuracy of the satellite. Firstly, Based on the scale factor on the ground test as the standard, the bias is calibrated. Secondly, the error model is reconstructed with the result of the first stage. The optimal method is utilized to recalibrate the bias and scale factor. In this paper, sun sensor and gyro are used to measure and correct the magnetometer error. Magnetometer error estimation method is derived, and satellite pitch angle and angular velocity are calculated through the pith filter using the output of sun sensor and gyro. Then, outputs of the filter are used to calculate magnetometer error. The estimated results are used to correct the magnetometer outputs through the filter. Simulation shows that the method is effective, and enhances the stability of the satellite. It is favorable for imaging tasks, and improves the micro satellite navigation system performance.