Method of Star Point Extraction for Daytime Infrared Star Image

The daytime infrared star image has the characteristics of high background noise, low signal-to-noise ratio and weak target, so it is difficult to accurately extract the star centroid by using the traditional single frame extraction method. Because the stars are weak and small targets, they are extremely sensitive to noise, and the star imaging is mostly random and irregular, so the robustness of extracting the star centroid by single frame is poor. Although the traditional multi-frame superposition method can overcome the problem of poor robustness of star centroid extraction in single frame, for infrared star images with high background noise, the superposition star image cannot significantly improve the signal-to-noise ratio, and the success rate of star extraction is still low. Therefore, this paper proposes a method that uses background prediction method to determine the position of suspected stars, expands the boundary, and then uses the expanded boundary to extract the energy of single frame star image, and superimposes the extracted energy star image to form a high signal-to-noise ratio star image, and finally extracts the star centroid. Experiments show that the accuracy of star extraction is 99.5%, which is 84.2% and 37.9% higher than that of adaptive threshold segmentation method and multi-frame superposition method, respectively, and 14.5% higher than that of background prediction method. At the same time, the accuracy of this method is improved by 12.8%, 41.4% and 33.3%, respectively, compared with the adaptive threshold segmentation method, background prediction method and multi-frame superposition method, which has obvious advantages over the traditional method.