Abstract: The scene degradation and ghosting in scene-based nonuniformity correction methods for infrared focal plane arrays were analyzed here, and a novel approach based on Gaussian filtering with edge constraint was proposed for nonuniformity correction of infrared focal plane arrays. A Gaussian filter with edge constraint was designed to obtain the desired estimation of the image. The gain and offset correction factors in the iterative formula were calculated using the steepest descent method. Meantime, the convergence speed was increased through adaptive control of the iterative step. Both simulation and experiments on real infrared images demonstrate that this approach can remove the fixed pattern noise more effectively compared with the existing methods. It can effectively suppress the scene degradation and ghosting, retain the detailed image information, and improve the image quality.