Applying SIFT Descriptors to Stellar Image Matching

Abstract

Stellar image matching allows to verify if a given pair of images belongs to the same stellar object/area, or knowing that they correspond to the same sky area, to verify if there are some changes between them due to an stellar event (supernova event, changes in the object position, etc). Some applications are stellar photometry, telescope tracking and pointing, robot telescopes, and sky monitoring. However, the matching of stellar images is a hard problem because normally the images are taken using different telescopes, image sensors and settings, as well as from different places, which produces variability in the image’s resolution, orientation, and field of view. In this context, the aim of this paper is to propose a robust SIFT-based wide baseline matching technique for stellar images. The proposed technique was evaluated in a new database composed by 100 pairs of galaxies, nebulas and star clusters images, achieving a true positive rate of 87% with a false positive rate of 1.7%.