Optical coherence tomography(OCT) has been shown to have advantages for obtaining images of biological tissues noninvasively and has great potential for applications in medical diagnosis.In this work,a method based on the empirical mode decomposition(EMD) method is adapted to improve OCT image quality by reducing the effects of noises.EMD is a time-frequency analysis,So it can more accurately analyze the time-varying non-stationary noise signal compared with the domain transform method.The experimental results show that EMD method can effectively suppress speckle noise,which improves signal-to-noise(SNR) to 7 dB and 3 dB for absence and presence of the scanner noise.It can also effectively remove the stripe noise due to instability of the scanning reference mirror in OCT as long as the EMD filter parameter is set properly.The fringe contrast can be reduced by more than 60%.The quality of OCT images is improved.Compared with the wavelet de-noising method,EMD has the advantages of simple operation steps,obvious de-noising effect,and the ability of effectively removing two kinds of noises.