固定阈值算法对由噪声引起的小的频谱系数进行归零,可以取得较好的去噪效果。但由于其阈值函数不连续,在阈值附近具有突变,所以会对一些大的噪声频谱系数有所保留,对小的信息频谱系数造成丢失。为了改善窗口傅里叶变换滤波算法中的阈值选取方法,提出了自适应阈值方法。通过模拟散斑干涉相位图验证可知,所提出的自适应阈值方法可以对噪声进行有效的滤除,同时对信号信息进行很好的保留。 The fixed threshold algorithm zeroes the small spectral coefficients caused by the noise, which can achieve good denoising effect. However, because of the discontinuous threshold function and the mutation near the threshold, some large noise spectral coefficients are reserved and the small information spectral coefficients are lost. In order to improve the threshold selection method in windowed Fourier transform filtering algorithm, an adaptive threshold method is proposed. By simulating the speckle interference phase diagram verification, we can see that the proposed adaptive threshold method can effectively filter out the noise and keep the signal information well.