【摘要】 电容层析成像图像重建是一不适定反问题。此种情况下,仅使用最小二乘法不能保证获得满意的介质分布图像重建结果,因此广泛使用TIkhonov正则化算法来产生适当的解。正则化参数的合适选取对图像重建至关重要,其对重建质量和计算时间都有影响。本文提出了一种基于最平坦斜率的Tikhonov正则化参数选择方法,并针对2种典型介质分布,将基于此方法计算的正则化参数同L-曲线法在电容测量数据无噪声和施加噪声情况下的图像重建结果进行了比较。更多还原

【Abstract】 Image reconstruction for electrical capacitance tomography is very often ill posed.In such cases, using only simple least-squares method can not ensure a successful image reconstruction of media distribu- tion.Therefore,Tikhonov regularization is widely employed to produce proper solutions.Correctly selec- ting regularization parameters is crucial for image reconstruction,which affects both quality and computing cost of the reconstruction.In this paper,a new parameter selecting method of Tikhonov regularization based on the flattest slope method is presented.The regularization parameters for two typical media distri- butions were calculated using both the proposed flattest slope method and the L-curve approach.The image reconstruction results for noiseless capacitance measurement data and for the data with noise added are given for comparison.更多还原