The influence of the detector energy resolution and the relatively close gamma energy emitted during nuclide decays often lead to peak overlapping of gamma spectra, which affects the accurate identification of nuclides. A novel method based on Gaussian convolution and curve fitting was proposed in this paper to achieve accurate identification of overlapping peak parameters in gamma energy spectra. The simulation results indicate that the maximum relative errors of peak position and peak width calculated by the novel method are 0.55% and 1.33%, respectively, which are smaller than the maximum relative deviations calculated by the particle swarm algorithm of 1.11% and 9.49% under varying Signal-to-Noise Ratio (SNR) levels. In unfolding the measured gamma spectra with 100-700 keV region, the maximum relative error between the proposed method and the Gamma Vision peak-seeking result is 0.10%, and the maximum relative error of peak width calculation is 2.34%. The experimental results demonstrated this method is feasible in the field of overlapping peak identification.