Electrical Resistance Tomography (ERT) is a kind of medical imaging technology taking the body's conductivity distribution as the target.With the advantages of visualization, low cost and no damage, it has broad prospects in the field of concrete defect detection. Based on the basic principle of resistance imaging, the MATLAB platform is used to compile the ERT forward and the inverse problem program. According to the electrical characteristics of concrete matrix and defects, we apply constraints to the algorithm and interfere with Newton-Raphson iteration process in inverse operation by modifying initial values, so as to improve the computational stability and image resolution of ERT posed inverse problem. The results show that the ERT constraint algorithm can reduce the number of iterations and eliminate imaging artifacts to a large extent, thus improving acuity and realizing relatively accurate nondestructive detection of concrete defects.