为提高车载激光多普勒测速低信噪比时的测速精度，同时保持较好的实时性，基于自相关检测算法在车载激光多普勒测速中的应用进行了理论与实验研究。从速度估计的克拉美罗下限（CRLB）出发，将自相关后信噪比与采样点数的关系和频域内信噪比估计相结合；不同原始信号的信噪比，可以选取最优重数和采样点数以便对信号进行自相关处理，从而减小计算量，实现系统信噪比和探测概率的提高。仿真和实验结果表明，在信噪比不低于-10dB时，二重自相关可以使系统速度估计精度接近CRLB。

In order to improve measurement precision of vehicularlaser Doppler velocity at low signaltonoise ratio(SNR) and keep a better realtime, the theoretical and experimental research based on the correlation detection algorithm was performed. Starting from the CRLB of speed estimation, the relationship of SNR with sampling points and SNR estimation in frequency domain was combined. The optimal autocrrelation multiplicity and sampling points were chosen according the SNR of the original signal, thus the calculation amount was decreased and the SNR and detection probability were improved. The simulation and experimental results show that double autocorrelation can make the estimation accuracy of the system speed basically reach CRLB when the SNR is not less than -10dB.