基于Otsu方法优化的ViBe日冕物质抛射检测

发布时间：2019-03-06 15:31

【摘要】：日冕物质抛射(Coronal Mass Ejection, CME)是太阳系内规模最大,程度最剧烈的能量释放过程,这些能量的释放可能对地球产生严重的影响,所以对日冕物质抛射的检测具有非常重要的意义。随着科技发展,自动检测CME的方法近年来被广泛的研究,本论文也在对日冕物质抛射的检测进行了研究,本论文的主要研究内容如下：(1)本论文介绍了目前现有的几种日冕物质抛射检测技术,例如基于霍夫变换检测CME的CACTus方法(the Computer Aided CME Tracking),基于亮度增强检测识别技术的SEEDS(the Solar Eruptive Event Detection)方法,以及人工检测技术；然后本论文对这三种方法进行了对比分析,其中重点将CACTus方法和人工检测方法进行了多角度的对比分析,可以得到人工检测虽然准确但是耗时较长,CACTus方法相较于人工检测容易出现误检漏检的问题。(2)为解决上述检测方法存在的问题,本论文将前景检测方法ViBe(Visual Background extractor)算法用于日冕物质抛射图像的CME检测,该算法的优点是能够很好地适应日冕物质抛射场景的变化,具有高的实时性和鲁棒性；缺点是在CME的检测过程中容易产生“鬼影”。针对这一问题提出采用Otsu方法(是一种自适应的阈值确定的方法,又叫大津法,简称Otsu)对ViBe算法得到的前景检测结果进行再次判别,将错判为前景点的点重新判断成背景点,从而得到更加准确的CME前景图像。然后对初步检测得到的CME前景检测结果进行极坐标变换,再根据CME的判定条件确定CME的个数和一些物理特性,同时根据实验结果制定出了CME目录。最后的实验部分通过本方法与CACTus方法和人工检测结果进行三个方面的对比,说明本论文方法在成功检测率上要高于CACTus方法,而误检率上要低于CACTus方法,同时能够检测到CACTus方法检测不到而人工却能检测到的微弱的CME运动目标。
[Abstract]:Coronal Mass Ejection (CME) is the largest and most intense energy release process in the solar system. The release of these energy may have a serious effect on the earth, so the detection of coronal mass ejections is of great importance. With the development of science and technology, the method of automatic detection of CME has been widely studied in recent years. for example, the computer-based CME Tracking method based on the Hough transform detection CME, the SEEDS (the Solar Erruptive Event Detection) method based on the brightness enhancement detection and identification technology, and the manual detection technology, and then the thesis compares the three methods, In that method, the CACTus method and the artificial detection method are mainly used for carrying out a multi-angle comparative analysis, and the problem that the manual detection is accurate but time-consuming is obtained, and the CACTus method is prone to misdetection and leakage detection compared with the manual detection. (2) In order to solve the problems existing in the above-mentioned detection method, the method of the future detection is used to detect the CME of the coronal mass ejection image, and the algorithm has the advantages that the change of the coronal mass ejection scene can be well adapted, and the method has high real-time performance and robustness; The disadvantage is that the "ghost" is easily generated during the detection of the CME. In view of this problem, an Otsu method (an adaptive threshold determination method is proposed, which is called Otsu), and the result of the foreground detection obtained by the ViBe algorithm is judged again, and the point of the misjudgment as the front scenic spot is re-judged to be a background point, So as to obtain a more accurate CME foreground image. And then the CME foreground detection result obtained by the preliminary detection is subjected to polar coordinate transformation, and the number and some physical characteristics of the CME are determined according to the judgment conditions of the CME, and a CME directory is developed according to the experimental results. In the last part, the method is compared with the CACTus method and the artificial detection result, and the method of the present thesis is higher than that of the CACTus method, and the error detection rate is lower than that of the CACTus method. At the same time, it is possible to detect a weak CME moving target that is not detected by the CACTus method but which can be detected manually.
【学位授予单位】：昆明理工大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P182.62

【相似文献】