According to the property that for Gaussian signals only,all cumulant spectral of order greater than two are identically zero,a Four-Order-Cumulant based Parameter Estimation (HOS-PE) approach was undertaken for the adaptive estimation of the gain and bias nonuniformity in infrared focal-plane array sensors from scene data. The statistical property of mean and variance for the gain parameter was given by mean of Taylor expansion and the analysis proves the estimation was approximately unbiased. The proposed method has been tested by the Monte Carlo simulation and verified. At last,HOS-PE method was tested through video sequences of real infrared data,reducing the fixed pattern noise and reaching high correction levels.