Comparative Study and Development of Two Contour-Based Image Segmentation Techniques for Coronal Hole Detection in Solar Images

Abstract

The study of solar coronal holes (CHs) is important in the understanding of solar physics and the prediction of space weather events, which have significant impact on space-based instruments, communication and navigation systems. With the availability of the multi-wavelength Atmospheric Imaging Assembly (AIA) instrument on board Solar Dynamics Observatory (SDO) satellite, a large volume of high-resolution solar images are produced continuously. Proper detection of CHs from AIA images is an important issue and recently, a few contour and machine learning-based techniques are found to be promising for such purpose. However, accuracy, time complexity and the requirement of human intervention are some of the critical issues with such methods. In this paper, to address these challenging issues, two contour-based approaches are developed, namely i) the Hough transformed simulated parameterized online region-based active contour method (POR-ACM) and ii) fast fuzzy c-means clustering followed by Hough transformed simulated static contour method (FFCM-SCM). The major issues that are addressed here are automated initialization of contour, reducing time complexity and avoidance of non-coronal hole inside a coronal hole region during contour evolution. The proposed techniques have been tested on three benchmark solar disk images and compared with the existing active contour without edge- (ACWE) based method and fuzzy energy-based dual contour method (FEDCM) of CHs segmentation. The results indicate the capability of the proposed techniques in detection and extraction of CHs in solar disk image with higher accuracy and reduced time complexity.

Appendix: Contour Initialization and Limb Correction

The elimination of solar limbs in the process of CHs region detection is also a vital problem. The presence of solar limbs leads to the misclassification of the CHs region. The stated issue has been handled by introducing the circular Hough transform and the corresponding outcome is shown in Figures 11 and 12.

1.1 A.1 Contour Initialization in POR-ACM

An experiment has also been conducted for different contour initialization and using different ACM under various conditions. Among ACM, here mainly POR-ACM has been considered for its fastness in segmenting the image. After initializing the contour near the object boundary of the solar image, as shown in Figure 10b, then implementing POR-ACM without eliminating off-disk pixels for segmenting out CHs, it has been found that the method fails for the purpose. The method plotted the contour on the boundary of the CHs region and on the boundary of solar disk along with the portion of the solar limb. As such, regions outside the bigger contour having low-intensity value also have been selected as the CHs regions (as shown in Figure 10c). But as a piece of prior information, it is to be considered as a non-CH region. Also, in the upper portion of the solar disk in the image located in the first row of Figure 10c it can be seen that the low-intensity region has been considered as the background. This is due to the fact that when POR-ACM forces the contour to move towards the boundary of the RoI, it considers the intensity homogeneity information for evolving the contour, and in this case, the intensities of CHs region and solar image background are homogeneous. The same output can be found for the contour initialized within the solar disk and inside the RoI, as shown in Figures 10d–10g.

Figure 10

Different contour initialization and corresponding outputs obtained using POR-ACM without eliminating off-disk region have been shown, where original images are given in (a), contour initialization near image boundary shown in (b), and corresponding outputs have been shown in (c), contour initialization inside the solar disk has been displayed in (d) and corresponding outputs are given in (e), (f) show the contour set-up within the RoI and related outcomes are displayed in (g).

Now, after initializing the circular contour using the Hough transform (shown in Figure 11a), then applying POR-ACM without eliminating solar-limb and image background will generate the same results as shown in Figure 10c. This is due to the presence of the off-disk region (solar-limb and solar image background). Now, eliminating the off-disk region after initializing contour, as highlighted in Figure 11d, then again applying POR-ACM has failed to reach the boundary of the CHs. The corresponding results can be seen in Figure 11e. This is due to the property of ACM, which finds the minima in a local manner and as such the contour gets stuck on the edge of the solar disk. The force at the solar disk edge is high due to the huge change in intensity value at the edge of solar disk. This force is preventing the contour to move further towards the CHs regions present inside the solar disk. However, after the application of the proposed POR-ACM, it can be found that the contour is able to reach the boundary of the CHs region as highlighted in Figures 2d and 11f. This is because the parameters in the introduced POR-ACM provide an extra force that overcomes the force extort by the change of intensity values at the edge of solar disk.

Figure 11

Various stages of POR-ACM-based CHs segmentation and contour initialization, where (a) original image, (b) initial contour, (c) results without eliminating off-disk region (d) elimination of off-disk region (e) output using OR-ACM after elimination and (f) final output using POR-ACM after elimination have been shown.

1.2 A.2 Contour Initialization in FFCM-SCM

From the second column of Figure 12, it can be seen easily that without the application of the solar-limb correction technique lead to the misclassification of the CHs region. The non-coronal holes regions present outside the solar disk region are also considered as CHs. Meanwhile, in the fourth column of the same figure it can be noticed easily that after the application of SCM employing a circular Hough transform based on the original image (as shown in Figure 12c) one has eliminated the problem of misclassification due to the presence of the off-disk region and has detected only the CHs region. This can also be established by the final outcome of the proposed technique shown in the last column of Figure 12. Basically, after the application of SCM the off-disk region is transformed into background or non-CHs region as given in Figure 12d. Thus, during this stage, the total image is left with only CHs regions (foreground) and non-CHs regions (background). As a consequence, the boundary of the CHs regions gets detected in the image. The final result can be seen in Figures 2e and 12e.

Figure 12

Various stages of FFCM-SCM-based CHs segmentation, where (a) original image, (b) output using FFCM, (c) static contour generation on FFCM results (d) Limb elimination and (e) final output have been shown.