Sunspot detection and tracking using image processing techniques

Abstract

Over the past five decades, space weather forecasting has gained significant interest since mankind started exploring the solar system using highly developed technology. Power and communication systems are critical for modern society to function optimally. However, these systems are often interrupted by natural events, such as solar activity. Examples of these solar events are solar ares and coronal mass ejections. Space weather research has taken a major role in the prediction of solar activity, especially since these events are the cause of interruptions of wireless communications systems, railway signalling, high voltage power transmission grids, and satellites. Systems that can detect and track sunspots, as manifestations of solar activity, have become frequent over the past few years. The images used in this study to detect sunspots were downloaded from the National Aeronautics and Space Administration did not need any changes to be made to the images and could be used immediately for implementation in this study. Three object detection and one object tracking algorithms were implemented for the detection and tracking of sunspots in this study. The first two object detection algorithms were both implemented and tested on the same image and then applied to different images to test the validity of the algorithms. The first algorithm's detection was based on finding the edges of sunspots and the second was based on finding segments in the image that resemble one of the nine chosen templates. The third algorithm was implemented on the extracted results of the first algorithm. The overall results obtained from the study show that the implemented algorithms prove to be accurate in comparison to the other studies' results found in Chapter 5. However, the definition of accuracy is different for detection and tracking and is discussed with the results. For future work, other traditional image processing techniques may be implemented to improve the results.