电容层析成像图像重建中的迭代算法

Iterative methods in ECT image reconstruction

电容层析成像技术中图像重建算法的准确与快速是其在工业实际中得以应用的关键。Landweber方法是一种简单、灵活的迭代算法,具有很好的正则化特性,它的投影变体能够提供非负的图像重建。Landweber算法实际应用的困难是其收敛速度太低。本文采用预处理方法来加快Landweber迭代方法的收敛速度,即通过较少的迭代次数,获得适当的重建结果。仿真实验表明,同经典的快速算法共轭梯度迭代方法相比,预处理Landweber迭代方法具有更好的重建图像质量。

The speediness and accuracy of image reconstruction algorithm in electrical capacitance tomography are the keys to its application in industries. Landweber method is a simple and flexible iterative algorithm and has very good regularizing features. Its projection variant provides nonnegative image reconstructions. The difficulty of Landweber method in practical applications is it converges too slowly. To overcome this difficulty, a preconditioning method is adopted to accelerate its convergence speed. That is, reasonable reconstruction result is obtained with less numbers of iterations. The proposed method features more competitive and robust than conjugate gradient iteration method that is fast but sometimes leads to lower accuracy.