Research On Automatic Detection Method Of Solar Radio Burst Based Using Otsu Binarization

Sun is the closest star to the earth,and it is also a very active celestial body.The study of solar radio bursts can not only reveal the laws of space weather,but also be closely related to our production and life.The phenomena of intense flux enhancement observed in the radio band is called solar radio bursts.Artificial detection method is still widely used in the observation and recording of solar radio bursts.This method is time-consuming and laborious,and has many disadvantages such as randomness,omission and so on.With the development of digital image processing technology,more and more methods such as denoising and parameter quantization are applied to radio astronomy.Because the formats of solar radio data recorded by various observatories are not identical,there will be great errors when the corresponding methods are directly applied,and there is still a certain gap in the real-time processing of the current automatic detection methods of solar radio bursts.On the basis of previous research results,combining image denoising,image segmentation,morphological operation and other image processing techniques,this thesis proposes an automatic detection method of solar radio burst suitable for the data recorded by Yunnan Observatories of Chinese Academy of Sciences.The automatic detection of type III radio burst is realized.Firstly,the format of DAT data recorded by Yunnan Observatories is converted into image format which is easy to process by Matlab.In the image denoising step,background subtraction method and channel normalization method are used for different types of noise to suppress the gradual background noise and horizontal strip interference in the image.By comparing the results of segmentation,the OTSU method,which has the most obvious effect on foreground and background segmentation and is most conducive to subsequent processing,is selected.In order to solve the problem of isolated noise in the segmented spectrum image,open operation is used to remove the isolated small points in the image,closed operation is used to bridge some small cracks in the image,thus the smoothing of the spectrum image is realized.For the spectrum image after morphological operation,the method of detecting straight lines and setting are used respectively.The threshold method quantifies the burst parameters.The reason why the data of Yunnan Observatories are not applicable by using Hough transform to detect straight lines is analyzed.Finally,by setting up two groups of comparative experiments,it shows that the automatic detection method of solar radio proposed in this thesis is feasible in detecting type III bursts.