In Internet of Things (IoT), radio frequency fingerprints (RFF) technology has been widely used for passive security authentication to identify the special emitter. However, few works took advantage of independent oscillator distortions at the receiver side, and no work has yet considered filtering receiver distortions. In this paper, we investigate the RFF identification (RFFI) involving unknown receiver distortions, where the phase noise caused by each antenna oscillator is independent. Three RFF schemes are proposed according to the number of receiving antennas. When the number is small, the Mutual Information Weighting Scheme (MIWS) is developed by calculating the weighted voting of RFFI result at each antenna; when the number is moderate, the Distortions Filtering Scheme (DFS) is developed by filtering out the channel noise and receiver distortions; when the number is large enough, the Group-Distortions Filtering and Weighting Scheme (GDFWS) is developed, which integrates the advantages of MIWS and DFS. Furthermore, the ability of DFS to filter out the channel noise and receiver distortions is theoretically analyzed at a specific confidence level. Experiments are provided when both channel noise and receiver distortions exist, which verify the effectiveness and robustness of the proposed schemes.