In the study of electronic stethoscope, analysis of phonocardiogram depends on the localization and identification of the first cardiac sound (CS) S1 and the second CS S2, which are basic components of the CS signal. However, in realistic environments the recorded CS signal is often mixed with the respiratory sound (RS) and other ambient interferences, which may cause failure in CS component localization and identification. In this paper, a CS localization and identification method is proposed, which is based on two steps: the rough CS localization step and the identification and mending step. The rough CS localization step involves calculating the approximate Shannon entropy, to roughly localize the CS components in low computational complexity. On account of the influence of RS and other ambient interferences, the aforementioned first step may fail to localize some CS components. By calculating and comparing the ratio of low-frequency power to high-frequency power for each CS component, the second step of the method can amend the rough CS localization results and identify which type of CS component they belong to. At last, the estimation of heart rate (HR) is also easily derived from the CS localization and identification results. Experiments using data recorded in various conditions shows the efficiency of the proposed method.