This paper presents a template matching-based blind recognition method for MQAM modulation. The method begins with estimating the carrier frequency, bandwidth, and code rate, followed by completing preprocess. Subsequently, it defines the constellation distribution entropy to describe the convergence degree of the constellation diagram distribution and achieves optimal sequence acquisition and residual carrier compensation through optimizing constellation distribution entropy. This process leads to the restoration of the optimal constellation diagram distribution. Finally, it calculates the matching degree between the optimal constellation diagram distribution and pre-set templates for different modulation types in order to accomplish modulation style recognition, including 16QAM, 32QAM, 64QAM, 128QAM, and 256QAM. This method is insensitive to initial parameter selection and leverages essential differences in constellation diagram distributions for robust recognition. Simulation results indicate that at a signal-to-noise ratio (SNR) of 15 dB, the recognition rate surpass 93%. Additionally, the proposed method exhibits significantly improved noise resistance compared to existing algorithms-thus validating its effectiveness.