Electrical Resistivity Tomography (ERT) is a non-invasive imaging technique to reconstruct the conductivity distribution of a field-sensitive field by measuring the field boundary voltage, thus realizing the monitoring of the interior of the sensitive field. An improved L-BFGS imaging algorithm is proposed to address the slow imaging speed and low accuracy of existing ERT imaging algorithms when dealing with large-scale problems. During the iteration process of the algorithm, we adopt a double-loop algorithm and a backtracking line search method, combined with the strategy of retaining only the information of the last several iterations, which significantly reduces the data storage requirement and the computational complexity. As verified by simulation experiments, the method effectively improves the quality of ERT images and shortens the time required for imaging.