Abstract: The all-day camera cloud image has important value in the field of astronomical observation, but the noise contained in the cloud image will have an adverse effect. The objective of this paper is to perform cloud image denoising processing.This paper has analyzed starlight noise, moonlight noise, daylight noise, lightning noise, reflective noise and special noise, and has summarized the characteristics of different types of noise distribution, region size and so on. Finally, eight kinds of image processing denoising algorithms such as boundary threshold method, timeline method and inverse proportional linear transformation are proposed based on the relative positional relationship with the cloud layer. The algorithms are combined to form a complete denoising system. Through experiments, different algorithms are compared to obtain the best denoising scheme under different noise conditions. According to the experimental results, we can draw conclusions that the system can effectively remove common noise, basically maintain the information integrity of the cloud layer and the sky, and can easily realize the cloud image denoising.