Aiming at the shortcomings of inaccuracy of estimation of the truncated singular value decomposition（ TSVD） algorithm to solution and the standard Tikhonov regularization is too smooth in the image reconstruction process of the electrical capacitance tomography system

based on regularization model

an improved TSVD regularization algorithm is proposed. The algorithm cuts the sensitivity matrix singular value into two parts

smaller singular value and larger singular value

by selecting the appropriate truncation parameter k

and then uses the Tikhonov-Gaussian filter function as the filtering factor

which not only corrects small singular value

but also overcomes the problem of excessive smoothing caused by the standard Tikhonov regularization due to the correction of large singular values. A variety of stream type simulations are performed on the improved algorithm using image reconstruction quality evaluation criteria. The simulation results show that compared with TSVD algorithm

the proposed algorithm reduces relative error by 26 % and increases correlation coefficient by 24 %

which has a good imaging effect.