Filtering in the Time-Frequency domain for the detection of compact objects

In this work we propose a detection method for compact sources in the Time-Frequency domain. By exploring the joint Time-Frequency distribution of data and by calculating the time-frequency profile of the compact sources it is possible to improve the detectability of faint point sources. We propose the detection of sources by identification of local maxima (peaks) in the time-frequency Wigner-Ville distribution of the data, previously filtered with a two-dimensional correlator filter. The filter serves the double purpose of enhancing the signal and of reducing cross-terms derived from the Wigner-Ville transform. We test our method and compare it with a the case of the detection in the time domain (filtered with the appropriate one-dimensional correlator), for a situation in which we have faint sources embedded in stationary white Gaussian noise. The detection in the time-frequency domain gives us better significance levels of faint sources (signal-to-noise ratio ~ 1) than the detection in just the time domain.