用光学干涉计量方法获得光场相位，通常都需要做去噪运算。在光场相位空间变化频率不高的情况下，目前的各种算法都能取得不错的效果。如果光场相位空间变化剧烈，传统去噪算法常引起细节丢失、条纹断裂。针对这个问题，通过引入光场的横向剪切，显著降低光场相位的空间变化频率，从而将噪声从光场中分离出来。给出了一种新的去噪算法和相应的理论分析，通过模拟计算和实验验证，证明该算法是可行和有效的，在光场相位空间变化频率较高的情况下，能得到比传统去噪算法更好的结果。

Denoising operation is usually needed for obtaining the phase of optical field by optical interferometry, and current denoise algorithms can achieve good results when the phase of optical field changes slowly. If the phase has high spatial variation frequency, the conventional denoise algorithms often cause loss of phase detail and fringe breaks. For this reason, by introducing shearing of optical field to reduce the frequency significantly and separating the speckle noise from optical field, a new denoise algorithm and corresponding theoretical analysis are presented. The validity of this new algorithm is showed in one simulated result and one experimental example of denoising; and the obtained phase approaches experimental value well, proved that this new algorithm is better than conventional algorithms.