In order to solve the problem that the original signal collected by Φ-OTDR optical fiber sensor contains a lot of noise and low signal-to-noise ratio, a denoising algorithm based on empirical mode decomposition and interval iterative invariant threshold is proposed. In this method, the signal is decomposed based on the time scale features,without setting the basis function in advance. And then a standard Butterworth filter is used to eliminate the noise in the intrinsic model components. Simulation experiments are designed, and the results show that the signal-to-noise ratio of humman beating signal and mechanical excavation signal is improved by 3.01 dB and 5.12 dB respectively, which can effectively restrain the noise of original signal data, thus improving the sensitivity of Φ-OTDR system.