随着数字全息术的应用日渐广泛，如何得到高质量的再现像质已成了数字全息研究和应用中非常重要的问题。由于散斑噪声的统计特性较差，因而散斑噪声被认为是相干成像中最难解决的问题之一。在研究数字全息术再现像散斑形成原因的基础上，利用激光散斑的统计特性，提出了降低再现像散斑的方法。先提取全息图再现像频谱中能量相对较小的频谱作为近似散斑频谱，并在该频谱段内进行数据随机替换，获取带有不同独立散斑的再现像, 在进行独立再现像散斑迭代叠加的同时逐步减少散斑能量，从而大幅减小散斑噪声，提高了再现像的像质。理论与实验均证明了其有效性,为数字全息图散斑噪声的消除提供了一种崭新的思路与方法。

With the increasing application of digital hologram, how to acquire high quality of reconstructed image of digital hologram has been a very important research topic. Speckle noise has been one of the most difficult problems in the coherent imaging system because of its poor statistical regularity. Based on the statistical characteristics and the formation cause of speckle noise in the reconstructed image of digital hologram, the way to reduce speckle noise is put forward. The frequency spectrum of reconstruction with lower energy is extracted from the reconstructed image of hologram and considered as approximate frequency spectrum of speckle noises. Then it is replaced with random data and the reconstructed image with different independent speckle noise is obtained. Lastly speckle energy is gradually reduced when the independent reproduced speckle noises are iteratively superimposed. As a result, the speckle noise can be reduced greatly and the quality of reconstructed image can be improved. The validity of this method is proved in theory and experiment. It offers a entirely new thinking and practical way to reduce the speckle noise in the reconstructed image of digital hologram.