To guarantee the preciseness and credibility of the epicardial potential mapping (EPM) signals used for cardiac electrophysiology research, pretreatment including denoising and characteristics recognition for EPM signals is necessary. So a new pretreatment method based on independent component analysis (ICA) and wavelet transform (WT) was designed: first to denoise EPM signals based on ICA with a new noise independent component selection measurement: the dispersion of independent components’ projection coefficients to every electrode, then to recognize the EPM signals’ characteristics according to their spline WT in the third scale. The results indicated that the method could denoise EPM signals efficiently with reserving or even strengthening the true depolarization signals detected at electrodes and recognize EPM signals’ characteristics correctly. So this new pretreatment method is fit for EPM system.