SPACE-SUIT: an Artificial Intelligence Based Chromospheric Feature Extractor and Classifier for SUIT

Abstract

The Solar Ultraviolet Imaging Telescope (SUIT) onboard Aditya-L1 is an imager that observes the solar photosphere and chromosphere through observations in the wavelength range of 200 – 400 nm. A comprehensive understanding of the plasma and thermodynamic properties of chromospheric and photospheric morphological structures requires a large sample statistical study of these regions, necessitating the development of automatic feature detection methods. To this end, we develop the feature detection algorithm SPACE-SUIT: Solar Phenomena Analysis and Classification using Enhanced vision techniques for SUIT, to detect and classify the solar chromospheric features to be observed from SUIT’s Mg II k filter. Specifically, we target plage regions, sunspots, filaments, and off-limb structures for detection using this algorithm. SPACE uses You Only Look Once (YOLO), a neural network-based model to identify regions of interest. We train and validate SPACE using mock-SUIT images developed from Interface Region Imaging Spectrometer (IRIS) full-disk mosaic images in Mg II k line, while we also perform detection on Level-1 SUIT data. SPACE achieves a precision of \(\approx 0.788\), recall of \(\approx 0.863\) and a MAP of \(\approx 0.874\) on the validation mock SUIT FITS dataset. Since our dataset is manually labeled, we perform ‘self-validation’ on the identified regions by defining statistical measures and Tamura features on the ground truth and predicted bounding boxes. We find the distributions of entropy, contrast, dissimilarity, and energy to show differences for the features in consideration. We find these differences to be captured qualitatively by the detected regions predicted by SPACE. Furthermore, we find these differences to also be qualitatively captured by the observed SUIT images, reflecting validation in the absence of a labeled ground truth. This work hence not only develops a chromospheric feature extractor, but it also demonstrates the effectiveness of statistical metrics and Tamura features in differentiating chromospheric features of interest, providing independent validation measures for any future detection and validation scheme.