Gerchberg-Saxton（GS）相位恢复算法被广泛用在多平面全息显示中，但是大部分平面的相关度很低。研究者们提出了一种补偿的方法，然而它并不有效。本文将这种补偿方法称作强补偿方法，并提出了一种基于弱补偿概念的方法。这个算法的核心是引进相应的权重因子改变进入算法的信息量从而实现弱补偿。数值模拟结果显示，同原始GS相位恢复三维算法相比，用这种算法各平面相关度的平均值和差值可以至少改善59.82%和97.03%。

Gerchberg-Saxton (GS) phase retrieval iteration algorithm is widely used in multi-plane reconstruction despite of the low correlation coefficient C values of most planes. Method of compensation, which we call strong compensation, was proposed by the researchers but it has little function. In this paper, an improved algorithm based on the concept of weak compensation is proposed. Introduction of weight factors to adjust the balance of the multi-plane reconstruction and to realize the weak compensation is the core concept in this proposal. The simulation results show that with this method the average and variation of C values can be improved by at least 59.82% and 97.03% compared with original GS iterative algorithm, respectively.