Period Estimation in Astronomical Time Series Using Slotted Correntropy

ECENT advances in photometric technologies have facilitated the proliferation of extensive astronomical surveys such as MACHO [1], OGLE [2], and recently Pan-STARRS [3]. A light curve is a time series in which the measured phenomenon corresponds to the brightness (magnitude or flux) of a stellar object. Light curves are the basic tool for the analysis of variable stars [4], whose brightness varies through time due to internal physical processes, or to external factors such as interactions with other astronomical objects. Some variable stars, such as eclipsing binaries (EB), cepheids, and RR Lyrae, exhibit periodic behaviors that are reflected on their corresponding light curves. For example, EB stars are systems composed of two stars, whose brightness shows periodic variations due to the mutual eclipses between them. The period of a light curve is a key parameter for classifying variable stars [5], [6], and estimating other parameters such as mass and distance to Earth [7]. Light curves are unevenly sampled due to constraints on the observation schedules: the day-night cycle, weather conditions, cali-Manuscript received February 02, 2011; revised March 28, 2011; accepted March 30, 2011.