Research On Dual-threshold Sunspot Segmentation Algorithm Based On Simulated Annealing Genetic Algorithm

Sunspots are important manifestations and main source of solar activities.Within sunspots,the darker cores are called the umbra and the peripheral,relatively bright regions are called the penumbra.Sunspots have a great influence on space weather and radio.Therefore,it is important to detect the umbra and penumbra of sunspots by a robust and reliable for forecasting space weather.There are many automatic methods for detecting sunspots.These methods involve various image processing techniques,all of which use the threshold method to segmenting the sunspots from the background.This methods have the following problems:(1)Using the a priori to estimate the intensity threshold,but as the solar magnetic field continues to evolve,the sunspots are constantly changing,and the fixed threshold detect the umbra and penumbra of sunspot is not accurate enough.(2)The sunspot on the edge of the sun cannot get a good detect is easily missed because of the limb-darkening.Thresholds play a decisive role detect the sunspots and the calculation of sunspots area.So how to get an accurate and adaptive threshold is very important.Based on the simulated annealing genetic(SAG)method,to self-adaptively derive dual thresholds for detecting the umbra and penumbra of sunspots simultaneously.The contents of the study are as follows:(1)Study to removed the limb-darkening,and preprocessing for the adaptive acquisition threshold of the SAG method.(2)Study the two thresholds of the umbra and penumbra of the detected sunspot using the SAG method.Focus on design selection,crossover and mutation operators.And a fitness function that fits the segmented sun umbra and penumbra.The genetic operator determines how quickly the population evolves to produce an optimal solution,and the fitness function determines whether an optimal solution can be found.(3)Study the SAG method set parameters,including:initial temperature,the cooling decay factor,the population size.(4)Comprehensive analysis and evaluation of the SAG method.(5)Improving the SAG method for image series.The research results are as follows:(1)We removed the limb-darkening by smooth average radial profile.The umbra and penumbra of sunspot are well segmented at the edge of the full-disk image.(2)Research and design of selection,crossover,mutation and fitness function in SAG.The discussion determined the initial temperature of the SAG method,which were10~5~10~6,0.999~0.9995,20~40.The two thresholds obtained by this method are very robust to the segmentation of sunspot umbra and penumbra.(3)Comparing the total sunspot areas,umbral areas,and penumbral areas obtained by the SAG method with the reported areas of HMIDD showed very good correlation.The correlation coefficients were up to 0.99,0.99,and0.95,respectively.The calculated sunspot area data has been published on the server for scholars to study.(4)The average computing time of an image by the improved SAG method was around one half that of the SAG method.