Scene-based statistical algorithms for nonuniformity correction assume the distribution of scene value to estimate the mean and variance of infrared energy received by each detector, then estimate the response parameters of each detector. In this way nonuniformity can be corrected. Current scene statistical algorithms for nonuniformity correction are analyzed. Interacting multiple model (IMM) is applied to correct nonuniformity. The experimental result shows that even taking consecutive frames as observing data, the nonuniformity can be corrected satisfactorily using IMM algorithm. It extends the applicability of Kalman filtering. More simulations for other algorithms have been done,and results indicate that the IMM algorithm has a better convergent performance.