基于改进总变差正则化算法的金属缺陷三维重建方法

为解决电磁层析成像(electromagnetic tomography,EMT)图像重建中的不适定性和病态性,将电磁层析成像用于金属缺陷检测,根据缺陷分布的稀疏性,提出了一种基于改进的总变差正则化算法(total variation,TV)的电磁层析成像图像重建方法,讨论了检测深度与激励频率的关系,利用三维重建算法对金属零件的表面和内部缺陷进行检测。通过仿真和实验评估了所提出算法的性能,并与Tikhonov正则化算法和L1正则化算法的重建图像和相对误差(relative error,RE)进行了比较。仿真和实验结果表明:使用改进的TV正则化算法重建的图像具有更好的图像重建效果和更小的相对误差,相对误差低至0.1左右,可以提高缺陷图像的重建质量和精度。

3D reconstruction of metal defects based on improved totalvariation regularization algorithm

In order to solve the ill-posedness and morbidity of electromagnetic tomography(EMT)image reconstruction,electromagnetic tomography is used for metal defect detection.According to the sparsity of defect distribution,an electromagnetic tomography image reconstruction method based on an improved Total Variation(TV)regularization algorithm is proposed.The relationship between the detection depth and the excitation frequency is discussed,three-dimensional reconstructed images are used for the surface and internal defects of the metal parts.The performance of the proposed algorithm was evaluated through simulations and experiments,and compared with the reconstructed image and relative error(RE)of the Tikhonov regularization algorithm and the L1 regularization algorithm.Both simulation and experimental results show that images reconstructed using the improved TV regularization algorithm have better image reconstruction effects and smaller REs,the relative error can be as low as about 0.1,which can effectively improve the quality and accuracy of the defects images.