A novel DFT-based approach for the estimation of the number of sinusoids

This paper proposes a method for the estimation of the number of sinusoidal components in a deterministic signal, based on an adaptive thresholding of the DFT The proposed method is described and compared with two common approaches present in literature, the Akaike information criterion (AIC) and the minimum description length (MDL). The proposal is also applied in the stage of the spectral estimation algorithm called IFFTc in order to detect hidden spectral components. The comparison is carried out for different signals and in terms of percentage of missed detections, percentage of false detections and elaboration times.