Research And Application Of Full-disk Solar Magnetic Field Image Registration Method

Image registration technology is a common method in image processing and pattern recognition.With the rise and development of artificial intelligence,it has been widely used in computer vision,medical image processing,image fusion,remote sensing images and other fields.There are differences in observation time,observation range,and observation equipment between different solar observing stations.For the convenience of scientific research,it is necessary to find a method for automatic registration and positioning of solar images between different observing stations.This paper studies the method of registration for the solar magnetic field image from the aspects of image preprocessing,image matching and registration,data analysis,accuracy assessment,and software design.According to the processing method of image information,the image registration method can be divided into the registration method of image gray information,the registration method of information in the image transform domain and the registration method of image feature information.Through the analysis of various methods,A method to match and locate solar magnetic field image based on scale-invariant feature point matches put forward.Firstly,pre-treatment for the original image is carried out by contrast enhancing and down sampling;Secondly,the feature scale-invariant feature points are extracted using scale-invariant feature point detection algorithm;Next,the feature points of the two images are matched and located roughly by seeking their homologous points;Finally,Based on coarse located area,the solar magnetic image can be matched and located exactly.Through the registration and positioning experiments of magnetic field images at different time periods,the parameters of the matching point logarithm,registration accuracy and registration error are quantitatively analyzed,and the tolerance analysis of the magnetic field images with different degrees of transformation is carried out.As proved by experiment results,the presented method can match and locate the solar magnetic images automatically,accurately and quickly.Finally,due to the need of observational data from astronomy researchers,a visualized solar magnetic field image registration software platform was implemented.It mainly realize step matching,automatic registration and batch registration function of solar magnetic field image,improve that speed and efficiency of registration,and has certain reference significance to other relevant fields.