Fast Fourier transform is a widely used method for signal processing in laser Doppler velocimetry systems. However, spectral leakage and fence effects occur in asynchronous sampling, and its processing accuracy is low. A hybrid convolution window based on the Nuttall window function and five-term maximum-sidelobe-decay window function is proposed to improve the detection accuracy of the six-spectral line interpolation correction algorithm. The hybrid convolution window can prevent the main lobe from becoming wide while ensuring good side lobe characteristics. The improved six-spectral line interpolation can effectively suppress the negative influence of the fence effect in the parameter estimation process and improve the analysis accuracy. The cubic B-spline interpolation is proposed to fit the interpolation coefficients to eliminate higher-order equations, and a frequency correction equation for the improved six-spectral line interpolation is derived. A dual-beam backscattering differential laser Doppler velocimetry platform is developed. The simulation data and measured signals demonstrate that the proposed algorithm exhibits good frequency and velocity measurement accuracy in the low signal-to-noise ratio environment.