Work

Cloud segmentation in astronomical images using gray level co-occurrence matrix and mini-batch k-means

Year: 2025

Type: article

Abstract: Night-time astronomical observations, which rely on ground-based optical telescopes with large field of view, are significantly affected by the clouds. Valuable information is not found solely in absolutely cloud-free sky regions, especially when tracking celestial targets. But cloud-affected images are typically discarded or otherwise poorly utilized, resulting in substantial waste of observational resources. To address this issue, a method of image segmentation is proposed as a means of extracting useful information from images affected by clouds. An objective image is divided into a number of regions, each exhibiting a different level of cloud interference, in order to extract valuable information. The gray level co-occurrence matrix is applied to the smoothed images, and the generated texture features are clustered combining mini-batch k -means with 8-connected component to classify regions into three segments: thick cloud, thin cloud and cloud-free ones. The efficacy of the proposed cloud segmentation is demonstrated by examining the magnitude-flux formulae of the sources. Furthermore, the robustness of the method is illustrated by statistical analysis of the O - C of the magnitudes in the presence of various perturbations.

Source: Fundamental Research

DOI: 10.1016/j.fmre.2025.12.019

Authors: Hao Fu, Houyuan Lin, Jianning Xiong

Institutions: University of Science and Technology of China, Chinese Academy of Sciences, Purple Mountain Observatory

Language: English

Topic: Image Retrieval and Classification Techniques

Subfield: Computer Vision and Pattern Recognition

Field: Computer Science

Domain: Physical Sciences

Open Access status: diamond

Funder: National Natural Science Foundation of China

Award: G4441998974