In this study, the matching pursuit method (MP) was used to decompose the second heart sound S2 into a series of time-frequency atoms. From these parameterized atoms, A2 and P2 can be separated. The first two dominant frequencies of A2 were identified and used as features of a linear classifier to diagnose aortic valve abnormality. This method was applied to two sets of S2 data recorded from 28 patients with normal, and 3-4 patients with abnormal, bioprosthetic aortic valves respectively, it was found that the values of both features exhibit significant differences between the normal and abnormal set (p< 5.0e-8). Using these two features, a correct classification of 90.3% was obtained. In addition, when the Wigner distribution of S2 was calculated from the decomposed atoms and compared with a spectrogram of S2, the MP method provided better results. The study demonstrates that the MP method may be promising technique for heart sound analysis.