A denoising method based on adaptive intrinsic time-scale decomposition (AITD) combined with the improved Bhattacharyya distance (IBD) and interval thresholding filtering (ITF) is employed for hemispherical resonant gyroscope in force-to-rebalance mode (FTR-HRG). The noise sources of the FTR-HRG are mainly the system noise (mechanical- thermal and circuit noise) and the environmental noise, and these are non-stationary noises. Therefore, the traditional filtering method is not effective in improving the gyro's noise characteristics. Firstly, the output of FTR-HRG is decomposed by AITD to obtain all proper rotational components (PRs). Secondly, the IBD method is used to calculate the probability density function (PDF) distance between the PRs and the raw data of FTR-HRG, and some of PRs with a large proportion of noise is filtered out. Finally, the PR signals with high signal content are filtered by interval soft threshold, and all processed signals are reconstructed to implement the filtering and noise reduction of FTR-HRG signals. The effectiveness of the proposed method is verified by typical Blocks signals and the decomposition filtering is performed on static and dynamic data of FTR-HRG, which verifies that the proposed method not only has the filtering but also has a short-time delay.