To eliminate the impact of the spoofing signal on a GNSS receiver, it is necessary to correctly identify the spoofing signal from the authentic signal. However, this is difficult as the two signals take after each other with almost the same characteristics. To improve the spoofing identification capability of the receiver, we propose a two-stage anti-spoofing processing method: The parameters of the two signals are estimated by maximum-likelihood estimation (MLE) at the base-band signal processing level, and the corresponding pseudo-ranges are grouped according to their relative amplitude relationship or/and code-carrier Doppler frequency coherence (CCDC) relationship. The grouping result is then confirmed or corrected by calculating the residual of the least-square estimation (including the pseudo-range residuals or carrier Doppler residuals) of the navigation solution derived from each group of signal components. If the grouping result is proven to be self-consistent, the group of the spoofing signals is identified using the residual of the LSE integrating pseudo-ranges from 1 or 2 low earth orbits (LEO)that are not attacked by the spoofer. Otherwise, regrouping is performed by trial-and-error method until the residuals are small enough. Once spoofing signals are identified after incorporating the LEO pseudo-ranges, the tracking loops are forced to lock on the authentic sides. The performance of the proposed method is tested by simulation, and the results verify the effectiveness of the proposed method for anti-spoofing signal processing. The receiver can keep reporting correct position, velocity, and timing (PVT) information in the case of success, complete failure, and partial failure of spoofing pull-off.