In this paper, a numerical scheme for simulation of DC potential drop (DCPD) signal is updated at first to deal with multiple defects in order to reconstruct multiple cavities in metallic foam (MF) from measured DCPD signals. A hybrid inversion strategy combining the genetic algorithm and the steepest descent optimization method is proposed then to evaluate the size and position parameters of the multiple defects. As example, simulated signals with artificial noise are adopted to reconstruct the profile of double cavity defects. The numerical results validated the hybrid inversion strategy and the updated forward scheme for sizing the multiple defects in metallic foam.