Wavelet transform is a popular tool for mechanical signal processing, while it suffers a few intrinsic defects which cannot be avoided within the scope of Mallat’s critically sampled filter bank. Redundancy is beneficial for enhancing the performance of wavelet transform as it allow higher design freedom for wavelet basis construction and improving the filter bank structure. Dual tree complex wavelet transform (DTCWT) is an discrete dyadic wavelet transform with a redundancy factor of 2. The wavelet basis of DTCWT basis is a hybrid one that organically integrates two related dyadic wavelet basis. DTCWT’s hybrid basis’ two wavelet functions form an pair of Hilbert transform pair, which equips the DTCWT with appealing properties of nearly analytic, nearly shift invariant and reduced spectral aliasing. These advantages of DTCWT are suitable for multi-component signal decomposition in mechanical vibration measurement. In this paper, a feature extraction technique based on DTCWT is proposed for rotating machinery’s vibration signal analysis. Numerical simulation and vibration signals from an experimental rotor setup are used to validate the effectiveness of proposed technique. It is shown that DTCWT show robust performance in analyzing multi-component signals, even in low SNR environments.