Novel 2-D real-valued sinusoidal signal frequencies estimation approach based on modified ESPRIT and propagator method

This paper considers the problem of estimating the frequencies of multiple 2-D real-valued sinusoidal signals, also known as Real X-Texture mode signals, in the presence of additive white Gaussian noise. Two algorithms for estimating the frequencies of real-valued 2-D sinusoids based on propagator method and modified ESPRIT method are developed. These techniques are direct methods, which do not require any peak search. A new data model for individual dimensions is proposed, which gives the dimension of the signal subspace is equal to the number of frequencies present in the observation. Then ESPRIT and propagator method based estimation techniques are applied on individual dimensions using the proposed new data model. Then an effective pairing procedure is used to pair the individual dimension frequencies. The performance of the proposed methods are demonstrated and validated through computer simulation.