Due to the influence of many physical factors of the neutron imaging system, the resulting neutron images usually suffer from severe noise pollution. Traditional PM-based methods cannot balance the noise removing and details preserving for neutron images. To address this problem, an improved partial differential equation method for neutron image denoising is proposed, in which the anisotropic diffusion properties are considered. The experimental results show that compared with the traditional PM and PM-based methods, the proposed method can effectively remove the Gaussian noise and achieve satisfying visual effect.