月球与深空探测技术研究室知识库

This paper proposes a method for extracting the solar centroid based on a multi-color model. The method is based on the three-color models (Lab, HIS, RGB) to segment and extract the color features of the image at the same time, coarsely locate the existence area of the sun target, and filter the obtained results through shape features, etc. We should try our best to optimize algorithm, for which can obtain a precise sun image. It is very important to combine circle center fitting with edge extraction to obtain the coordinate of edge point. What's more, we need to select the edge points to form a sample set of edge points. It is helpful to fit the center of the circle by the least square method. The algorithm is improved to find the center of mass of the sun. By subjective judgment and error analysis evaluation of a large number of results, this method is useful as we expecting for the segmentation and extraction of solar targets. But due to the interference of the shooting angle, light, exposure, weather and other factors when the image was taken, which makes the result of target segmentation not ideal. However, it has little effect on the final centroid extraction result. © 2021 SPIE.