To satisfy the accuracy of image reconstruction, this paper carries out offline optimization of the Hessian matrix in the modified Newton-Raphson algorithm (MNRA) for image reconstruction of electrical resistance tomography (ERT). Firstly, the selection strategy of regularization factor, which directly affects the accuracy of the reconstructed image, was discussed in details. Next, the improved particle swarm optimization (PSO) algorithm was adopted to alleviate the ill-posedness of the Hessian matrix through offline optimization. The variables of offline optimization include the radius ratio between each layer of the finiteelement model (FE model) to the sensitive field (SF) during the γ-refinement of the ERT, and the positions of the nodes added through element subdivision. The experimental results show that, under the same conditions, the above optimization measure can improve the solution accuracy of the ERT's inverse problem by alleviating the ill-posedness of the Hessian matrix, which is used to correct the dielectric resistance distribution (DRD) in the SF, without sacrificing the real-time performance of the MNRA.