In order to solve the contradiction between noise suppression and signal tracking in filtering for fiber optic gyroscope (FOG), a method as AMA-DWT-DMKF is proposed. Using the AMA algorithm, FOG output signal is divided into a signal transition region and a stability region, and combining the advantages of two algorithms such as DWT and Kalman with different Q/R values, in the end giving a targeted filtering for the results of the AMA algorithm. The proposed method is used for denoising the static signal, disturbance signal and the change rate signal. Experimental results show that the proposed algorithm not only has higher denoising capability, but also has a better signal tracking capability.