The Application Of Kalman Filtering In CME Image Processing

Coronal Mass Ejection(CME)are the largest scale and most active phenomenon of solar activities,which can bring serious effects on the earth environment.This phenomenon could cause disruption of communications,failure of space,breakdown of satellite,collapse of power grid and power equipment.Therefore,the extraction and detection of CME makes great significance to the prediction of space weather.In factor,the process of detection and extraction of CME depend on mathematical modeling methods and image processing technology from the image sequence.The detection of CME can be seen as modeling of backgrounds from sequence images,in order to detect the CME which consider as foreground.In the light of complex feature of the background environment(planets and comets,changes in light)for CME-projected sequence images,a modified background updating algorithm based on Kalman filter is proposed,which after comparing and analyzing the characteristics and disadvantages of the optical flow,frame difference and traditional background difference in this paper.The modified method based on the sequence image under polar coordinates,the dynamic background image is established by modifying the traditional model initialization and the model learning rate,then background difference and morphological technique are used in post-process in order to detect and identify the CME,at last,the CME is tracked.Compared with optical flow,frame difference and traditional background difference,the outline of CME detected by this method is relatively complete,the accuracy is higher,the error is smaller,and it can adapt to the complicated background environment in the solar sequence image.The modified method in changing dynamic environment can better remove noise,coronal flow and other disturbances,creating a more realistic background.The main research contents and results of this paper are as follows:(1)Description of the background and significance of CME research,the research methods of moving target detection,the current situation of CME research methods at home and abroad,and the difficulty of research on moving target detection.(2)The current mainstream frame difference method,optical flow,background difference method and traditional Kalman algorithm are introduced in this paper,and the advantages and disadvantages of the algorithm are compared and analyzed deeply.(3)In view of the main problems existing in the current moving target detection algorithm,the traditional Kalman filter background updating algorithm is improved,and the improvement process of the algorithm is introduced in detail.The main contents of the improvement include the initialization parameter estimation of the background model,the updating of the background model and the extraction and processing of the CME motion area in the postprocessing.And the improved algorithm is applied to the extraction of CME.The main steps include image preprocessing,improved Kalman background extraction,determination of moving material,connectivity analysis of moving target,hole filling,threshold segmentation and so on.(4)The modified Kalman background method in this paper is compared with the traditional algorithm.Based on the CDAW,the SEEDS,CACTus,frame difference method and the traditional Kalman filter background updating method are carried out respectively contrast and analysis.Experiments show that it can not only detect the CME in the LASCO C2 sequence image,but also detect the CME on the C3 sequence image that not detected on the CDAW manual directory.Compared with manual recognition,the detection method in this paper is faster and more robust.In terms of accuracy,it can not only detect the CME listed on the CDAW manual directory,but also can detect the relatively weak brightness of the CME,improve the accuracy of automatic detection algorithm.At the same time,the detection of the CME tracking effect makes high performance.But there is still a small false alarm rate.