Infrared focal plane arrays imaging system tend to be affected by the interior thermal radiation that drifts slowly over the time, resulting in the continuous spatial lowfrequency nonuniformity noise in the image. In this paper, a scene-based nonuniformity correction method for interior thermal radiation is proposed. The gradient domain least mean square (LMS) algorithm is used to estimate the gradient caused by the interior radiation noise according to the fixed spatial low-frequency information in multiframe images. After that, the calibration parameters are obtained by fitting the derivatives of the polynomial-based correction model with the calculated gradient image. The performance of the proposed method is thoroughly studied with infrared image sequences with simulated nonuniformity and real nonuniformity. The results show that the proposed algorithm can remove interior radiation noise more completely without ghosting artifacts than the state-of-art algorithms and is more robust to different scenarios.