We are focused on the phase retrieval and 3D profiles of microparticles (MP), analyzed in optical trapping. To this aim, we use several image segmentation procedures of MP from an optical trapping video. Moreover, we implement different procedures to remove several errors in the segmented image by means of algorithms based on the Legendre polynomials to enhance the MP image quality, as well as the errors caused by the drift inducing a shift in the MP (considering different MP with drift by shift, static, trapped and untrapped). Once the image corrections are conducted considering these errors, we propose some processes to recover the irradiance distribution of the MP in the image, from which it is possible to retrieve the MP phase to show their 3D profiles as 3D objects. Finally, we analyze the image in gray tones and use different retrieval methods with 2D Legendre polynomials, Gerchberg–Saxton (GS), Goldstein, and Quality Guided. We propose some modifications to such algorithms using Legendre polynomials, resulting in the deformation re-scaling iterative algorithm, ModI-LP, and a modified GS using the 2D-LP code to optimize our results.

我们专注于在光学捕获中分析的微粒 (MP) 的相位检索和 3D 轮廓。为此，我们使用了来自光学捕获视频的 MP 的几种图像分割程序。此外，我们通过基于勒让德多项式的算法实现不同的程序来消除分割图像中的几个错误，以提高 MP 图像质量，以及由引起 MP 偏移的漂移引起的错误（考虑不同的 MP 与通过移动、静态、捕获和未捕获的漂移）。一旦考虑到这些错误进行图像校正，我们提出了一些过程来恢复图像中 MP 的辐照度分布，从中可以检索 MP 相位以将其 3D 轮廓显示为 3D 对象。最后，我们以灰色调分析图像，并使用不同的检索方法与 2D Legendre 多项式、Gerchberg-Saxton (GS)、Goldstein 和 Quality Guided。我们建议使用 Legendre 多项式对此类算法进行一些修改，从而产生变形重新缩放迭代算法 ModI-LP 和使用 2D-LP 代码的修改后的 GS 以优化我们的结果。