利用四种常见算法对由球面参考光像面数字全息显微术得到的中药饮片海金沙细胞的包裹相位图进行了相位解包裹。结果表明：基于离散余弦的最小二乘算法和预条件共轭梯度算法会使欠采样区域的误差进行传递，使得解包裹相位产生较大误差; 质量图导向路径跟踪算法由于未识别残差点，导致出现“拉线”或者“孤岛”区域，而且处理速度很慢; 基于横向剪切干涉的最小二乘算法处理欠采样包裹相位图的效果最好，而且处理速度较快。

The existing phase-unwrapping algorithms are analyzed and classified, aiming at solving the problem of under-sampling in wrapped-phase map. The least-square algorithm based on lateral shearing interferometry is studied. Four common phase-unwrapping algorithms are applied to the wrapped-phase map of Lygodium japonicum cell, which is obtained by image-plane digital holography recorded with spherical reference waves. The results show that the least-square algorithm based on discrete cosine transform and the pre-condition conjugate grads algorithm result in transfer of phase-unwrapping error in the under-sampled area; the quality-guided path following algorithm results in pull phenomenon and island area because of the residual point without recognition, and its processing is slow; the least-square algorithm based on lateral shearing interferometry has the best effect for dealing with the under-sampled wrapped-phase map, and its processing is the fastest.