数字全息显微三维成像过程中,相位解包裹是三维重构中的关键技术。在Lp范数框架下,研究了二维相位解包裹算法统一的数学模型,分别运用了多种算法对数字全息显微实验得到的包裹相位图进行了实验分析。实验结果表明,L0范数方法总体上计算效率最高,但是由于噪声点过多而不能得到正确的结果;L1范数法能够获得较好的全局解,但是还是有一定方向上的畸变,并且耗时最多;最小Lp范数法,耗时较多,进行了两层的多次迭代,计算结果却有畸变;而最小二乘法和加权最小二乘法,计算效率较高,满足准实时测量的要求,并且得到比较理想的三维重构结果,是数字全息显微中可取的相位解包裹算法。

Phase unwrapping is a key step of three-dimensional display in digital holographic microscopy(DHM). On the basis of the framework of Lp-norm,the unified mathematic model of two-dimensional phase unwrapping is studied. A wrapped phase map of DHM is analyzed with several phase unwrapping methods. The results show that the L0-norm algorithm performed best in efficiency,but got wrong result because of too much noise,the L1-norm algorithm could achieve an better resolving,but the most long time consuming,the Lp-norm algorithm used much iterative,but also resulted with aberrations and the efficiency was rather low,the least-squares algorithm and the weighted least-squares algorithm could obtain better optimal resolving and better efficiency. The two algorithms are acceptable in DHM.