Digital Image Processing Research On Sun Hand-painted Sunspot Records

Sunspot is the earliest record of the solar activities.It has been systematically observed for about 400 years after the invention of telescopes.Study the long-term evolution of solar activities,especially its main 11 year solar cycle,has been obtained based on these data to a great extent.In recent years,these hand-drawing records of sunspots are being digitally converted for long-term preservation and use of computer technology to processing.With the progress of observation equipment and technology.The observation of the hand-drawing sunspots record will be inevitably replaced by the direct digital image of the CCD In the future.Therefore,the correlation analysis of the two kinds of observation records obtained by the coexistence of the two observation methods,especially research for the difference between the two types of observation records,this is very important for the convergence of traditional hand-drawing sunspots records with the future of pure digital sunspots records.In this context,based on the research of related image processing technologies,and emphatically analyzes the techniques and methods that used in the correlation analysis of hand-drawing records of sunspot and SDO/HMI(Solar Dynamics Observatory/Helioseismic and Magnetic Imager)images.On the basis of above,we make correlational research for the two types of sunspot records:(1)In order to more accurately analyze the accuracy of the hand-drawing records of sunspot,we selected samples from the Purple Mountain Observatory's hand-drawing records of sunspot,and design experiments to measure the scale error of the scanner which used in the scanning process,as well as the green solar ring in the recording paper with the printing accuracy problems that lead to its inconsistencies diameter length of east-west and north-south directions.And by extracting the sun circle line in four directions to measure the direction angle which comes from the process of scanning.The experimental results show that the scale of the scanner in the horizontal and vertical directions is 0.2%,the ring of the sun on the recording paper is not a perfect circle,and the diameter in the east-west direction is 1%shorter than that in the north-south direction,the orientation error of the record paper can reach up to 0.5 degree in scanning.(2)In order to identify the sunspots in the hand-drawing records,We design the experiment and use the maximum color difference method to respectively extract the background color of the paper,and extract the gray pencil handwriting extraction in the hand-drawing records of sunspot according to the distribution characteristics of different color in three color component values of R,G and B in the hand-drawing records of sunspot,and then compare the experimental results.The results showed that the effective extraction rate of background of the recording paper is about 57%,and the effective extraction rate of gray pencil handwriting is 97%.We first use multiple connected domain geometry and shape feature values combination filtering interference factors,and finally realize find and retain the sunspot region.Second,we designed the BP network that trained with MNIST database and apply it to the recognition of handwritten numbers on recording paper to remove more interference factors.Experimental results show that the BP network can recognize more handwritten numerals.However,due to special reasons,the false recognition rate is up to 6%,while the traditional method of misrecognition rate is 2%.(3)In order to realize the correlation analysis of sunspots in two kinds of images.First,we correct the P angle of the SDO/HMI,and then deduct limb darkening of full solar disk.We extract the local minimum value to obtain sunspot's position,and through the image reconstruction method to obtain sunspot's area,and finally realize to identify sunspots.Second,we correct the various errors,direction angle,P angle of the hand-drawing record image,and adjust its pixel resolution to be consistent with the SDO/HMI image.After these processe is complete,we comparing the sunspot position of hand-drawing records with the simultaneous SDO/HMI images of continuous spectrum by overlapping method to find out the correspondence between the two types of sunspots.We Statistics its coordinate position,size,area,and use histogram,Gaussian curve fitting,Rayleigh distribution to display and make statistical analysis of the results.The experimental results show that the mean of the two types of sunspots recorded in the direction of latitude deviation is 0.94 arc seconds(mm 0.08);the standard deviation of the positional deviation of the two types of sunspots is 5.3 arc seconds(0.47 mm).The mean of the two types of sunspots recorded in the direction of longitude deviation is-2.2 seconds(-0.2 mm).The hand-drawing record observation accuracy is about 7 arcsec(0.6mm),the lower limit of the hand-drawing scale of hand-drawing sunspots is about 4 arc seconds(0.35mm),and the correlation coefficient of the equivalent radius of two kinds of sunspot records is-0.36.In this paper,through the correlation analysis of two types of sunspot records,we obtained the difference between the hand-drawing records of sunspot and the pure digital acquisition of the observed record that provides an important basis for the convergence of two types of sunspots.Based on the experimental results,we can correct the various errors in the hand-drawing records of sunspot that can provide a more accurate data of hand-drawing sunspot records for astronomers,and provides valuable and reliable data for the study of sunspots long-term evolution.