Subsea images measured by the side scan sonars (SSSs) are necessary visual data in the process of deep-sea exploration by using the autonomous underwater vehicles (AUVs). They could vividly reflect the topography of the seabed, but usually accompanied by complex and severe noise. This paper proposes a deep denoising method for SSS images without high-quality reference data, which uses one single noise SSS image to perform self-supervised denoising. Compared with the classical artificially designed filters, the deep denoising method shows obvious advantages. The denoising experiments are performed on the real seabed SSS images, and the results demonstrate that our proposed method could effectively reduce the noise on the SSS image while minimizing the image quality and detail loss.