Planar array capacitance imaging based on adaptive Kalman filter

Planar array capacitance imaging system has the characteristics of uneven distribution of sensitive field, serious ill posed problem and measurement data vulnerable to external interference, and these characteristics will make the image artifacts particularly serious, affect the quality of the reconstructed image, and even determine the number of defects with difficulty. In order to solve the problem that the edge electric field and ill conditioned characteristics of planar array electrode seriously affect the quality of capacitance image reconstruction, an improved image reconstruction algorithm based on adaptive Kalman filter is proposed to reduce the noise of capacitance data and dielectric constant matrix. On the basis of constructing the state model of planar array capacitance imaging with noise, the maximum likelihood criterion is used to estimate and modify the noise variance matrix of dielectric constant matrix on-line, and the noise variance matrix of dielectric constant matrix is modified in real time. In order to restrain the filtering divergence and accelerate the convergence speed, different weighting coefficients are provided for the error covariance matrix with time going by. Through designing four kinds of samples from simple to complex structure, the defect detection experiment of composite structure is carried out. The experimental results show that compared with linear back projection (LBP), Tikhonov regularization (TR) algorithm and Kalman filtering algorithm, the image error of adaptive Kalman filtering algorithm can be reduced by about 20%, the image correlation coefficient is as high as 0.79 and the convergence speed can be improved by about 15%, the image artifacts of the four samples are greatly reduced. The experimental data show that the proposed adaptive Kalman filter image reconstruction algorithm can effectively reduce the noise of capacitance and permittivity matrix, enhance the stability of planar array capacitance imaging, and reduce the image error, so that the quality of the image can be significantly improved. This study provides a strong technical basis for improving the quantization accuracy of planar array capacitance imaging detection. In the future, we will further consider the image reconstruction under the condition of complex object field.