Aiming at the problem of traditional wavelet function in processing laser gyroscope output signal, a new wavelet threshold denoising method with parameter threshold function, adaptive determination of optimal decomposition layer number and optimal threshold is proposed. First, a new adaptive thresholding function is proposed. Then, the wavelet proportional energy entropy is calculated based on the principle of maximum energy entropy, and the optimal number of decomposition levels of the wavelet is adaptively determined, and the combination method of SURE (Stein Unbiased Risk Estimator) unbiased estimation principle and Newton iteration method is used to determine the optimal threshold of signal change with time adaptively. Finally, the experimental verification is carried out by using the measured data and Allan variance analysis. Experimental results show that both static laser gyroscope signal and dynamic laser strapdown imu signal, the improved adaptive wavelet noise reduction method of the noise reduction result is better than that of the traditional wavelet thresholding method and the standard Kalman filtering method, and the method processed signal higher precision, smaller mean square error and smaller noise coefficient, which effectively restrain the interference of noise of laser gyroscope output signal.