Automated tracing of open-field coronal structures for an optimized large-scale magnetic field reconstruction

Solar Probe Plus and Solar Orbiter will provide detailed measurements in the inner heliosphere magnetically connected with the topologically complex and eruptive solar corona. Interpretation of these measurements will require accurate reconstruction of the large-scale coronal magnetic field. In a related presentation by S. Jones et al., we argue that such reconstruction can be performed using photospheric extrapolation methods constrained by white-light coronagraph images. Here, we present the image-processing component of this project dealing with an automated segmentation of fan-like coronal loop structures. In contrast to the existing segmentation codes designed for detecting small-scale closed loops in the vicinity of active regions, we focus on the large-scale geometry of the open-field coronal features observed at significant radial distances from the solar surface. The coronagraph images used for the loop segmentation are transformed into a polar coordinate system and undergo radial detrending and initial noise reduction. The preprocessed images are subject to an adaptive second order differentiation combining radial and azimuthal directions. An adjustable thresholding technique is applied to identify candidate coronagraph features associated with the large-scale coronal field. A blob detection algorithm is used to extract valid features and discard noisy data pixels. The obtained features are interpolated using higher-order polynomials which are used to derive empirical directional constraints for magnetic field extrapolation procedures based on photospheric magnetograms.