基于TV正则化算法的电容层析成像自适应剖分方法  

Self-adaptive mesh generation in electrical capacitance tomography based on total variation regularization algorithm

作　　者：王化祥[1] 唐磊[1] 

机构地区：[1]天津大学电气与自动化工程学院,天津300072

出　　处：《电子测量技术》2006年第5期8-11,共4页Electronic Measurement Technology

基　　金：国家自然科学基金(60532020;60301008;50337020;60472077);国家高技术研究发展计划专项经费(20001AA413210)资助项目

摘　　要：针对电容层析成像逆问题解的不适定性及重建图像分辨率较低的问题,提出一种基于总变差(totalvariation,TV)正则化的图像重建算法,并由该算法提出一种自适应剖分方法。同常见的Tikhonov正则化算法相比,这种新算法不仅保证了逆问题求解的稳定性,而且提高了对介质非连续分布区域成像的分辨能力,具有良好的保边缘性。基于该算法的自适应剖分方法能够根据介质分布自动对剖分网格进行局部细分。相比全区域细分方法,这种剖分方法在保证图像分辨率的同时减少了计算量,提高了图像重建速度。实验结果表明,该算法在重建图像质量和实时性两方面均具有优势。To solve ill-posed problem and poor resolution in electrical capacitance tomography, a new image reconstruction algorithm based on total variation regularization is proposed and a new self-adaptive mesh refinement strategy is put forward. Compared with Tikhonov regularization which is commonly used, this new algorithm not only stabilizes the reconstruction, but also enhances the distinguish ability of the reconstruction image in areas with discontinuous medium distribution. It possesses a good edge-preserving property. The self-adaptive mesh generation technique based on this algorithm can refine the mesh automatically in specific areas according to medium distribution. This strategy keeps high resolution as mesh refinement all over the region but reduces calculation loads, thereby speeds up the reconstruction. Experimental results show that this algorithm has advantages in terms of the resolution and realtime performance.

关 键 词：电容层析成像 图像重建算法 TIKHONOV正则化 总变差 保边缘性 自适应剖分 

分 类 号：TP271.61[自动化与计算机技术—检测技术与自动化装置]