Abstract
To overcome limitations that occur in time-frequency representations (TFR) of signals, a new technique that considers the ambiguity function (AF) is proposed. This work firstly introduces a filtering technique that is based on the processing of the image of the AF of the signal. Secondly it shows the potential of this technique in application to ECG signal analysis by comparing some kernel-based techniques. The analytic form of the ECG is used to eliminate aliasing and the cross terms generated by the negative spectra that appear in TFRs. This work shows that data adaptive time-frequency analysis of the ECG by image processing techniques is possible and potentially useful.