Kostenko Alexander, Batenburg K Joost, Suhonen Heikki, Offerman S Erik, van Vliet Lucas J
Department of Imaging Science & Technology, Delft University of Technology, Delft, The Netherlands.
Opt Express. 2013 Jan 14;21(1):710-23. doi: 10.1364/OE.21.000710.
State-of-the-art techniques for phase retrieval in propagation based X-ray phase-contrast imaging are aiming to solve an underdetermined linear system of equations. They commonly employ Tikhonov regularization - an L2-norm regularized deconvolution scheme - despite some of its limitations. We present a novel approach to phase retrieval based on Total Variation (TV) minimization. We incorporated TV minimization for deconvolution in phase retrieval using a variety of the most common linear phase-contrast models. The results of our TV minimization was compared with Tikhonov regularized deconvolution on simulated as well as experimental data. The presented method was shown to deliver improved accuracy in reconstructions based on a single distance as well as multiple distance phase-contrast images corrupted by noise and hampered by errors due to nonlinear imaging effects.
基于传播的X射线相衬成像中用于相位恢复的先进技术旨在解决一个欠定线性方程组。尽管存在一些局限性，但它们通常采用蒂霍诺夫正则化——一种L2范数正则化反卷积方案。我们提出了一种基于总变分（TV）最小化的相位恢复新方法。我们使用各种最常见的线性相衬模型，将TV最小化用于相位恢复中的反卷积。我们将TV最小化的结果与蒂霍诺夫正则化反卷积在模拟数据和实验数据上进行了比较。结果表明，对于因噪声而损坏且受非线性成像效应导致的误差影响的基于单距离以及多距离相衬图像的重建，所提出的方法能提供更高的精度。
