Wang Chenxing, Da Feipeng
Appl Opt. 2014 Sep 20;53(27):6222-9. doi: 10.1364/AO.53.006222.
In dynamic 3D measurements, the recovering carrier signal as well as phase retrieval are important issues for single fringe image analysis. Local mean decomposition is a powerful signal demodulation tool, but it usually encounters an obstacle, namely, the mode-mixing problem. Then different components, especially noise and carrier signals, are all probably mixed in one of the decomposition results, thus confusing its physical meaning. Utilizing the characteristics of original noise, we design a pair of differential signals based on the conditions that two mixed components should meet to be separated completely and then add them to the original signal. Re-decomposing the newly formed signal, the differential signal, along with the original noise, will be separated from the carrier signal, leaving very little negative impact due to the characteristics of the same amplitude and opposite polarity of the differential signal. With the mode-mixing problem of high-frequency components being resolved, the decomposition of the following low-frequency components becomes more reasonable, facilitating the fringe pattern analysis and further phase retrieval. The proposed method is suitably used for the signal, even though it is not stable. Experiments illustrate the efficiency of this novel adaptive method.
在动态三维测量中，恢复载波信号以及相位检索是单条纹图像分析的重要问题。局部均值分解是一种强大的信号解调工具，但它通常会遇到一个障碍，即模式混合问题。然后，不同的分量，尤其是噪声和载波信号，都可能混合在其中一个分解结果中，从而混淆其物理意义。利用原始噪声的特性，我们根据两个混合分量要完全分离应满足的条件设计了一对差分信号，然后将它们添加到原始信号中。对新形成的信号（差分信号）进行重新分解，差分信号以及原始噪声将与载波信号分离，由于差分信号幅度相同、极性相反的特性，其负面影响很小。随着高频分量的模式混合问题得到解决，后续低频分量的分解变得更加合理，便于条纹图案分析和进一步的相位检索。所提出的方法适用于该信号，即使它不稳定。实验证明了这种新颖的自适应方法的有效性。
