Reconstruction of Non-stationary Signals with Missing Samples Using Time-frequency Filtering

This study proposes a new time-frequency (TF) method for the recovery of missing samples from multicomponent signals. This is achieved by a combination of a sparsity-aware TF signal analysis method with TF filtering technique. A sparsity-aware TF method overcomes distortions caused by missing samples in the TF domain. This is followed by the use of TF filtering techniques for recovery of signals. All the extracted components are then combined to recover the complete signal. The proposed method outperforms other signal recovery methods such as gradient descent algorithm and matching pursuit.