New Star Detector and Descriptor for Deep-sky Image Live Registration

Recently, a new image registration application named live stacking appears in deep-sky imaging fields. This kind of live registration mainly handles translation and rotation caused by vibration of imaging devices and the Earth's rotation. The existing star-based registration methods mainly have two defects: highly parameter related and very complicated matching mechanism to cope with image rotation instead of using rotation invariant descriptor. In order to make up these defects, in this work, a live registration algorithm via star detection and description is proposed. Firstly, an automatic detection method based on Otsu, centralization and contour tracing is proposed to implement star detection. Then, a translation and rotation invariant binary descriptor is constructed to serve feature matching. Extensive experiments illustrate that our proposed method is feasible in deep-sky image live registration compared with these state-of-the-art feature-based methods used in computer vision fields.