The identification of multiphase flow regimes is helpful to the further study of other problems in the multiphase flow field, and the efficient and safe operation of the actual production process. In order to identify the flow regimes of gas-liquid two-phase flow, a method of features fusion based on the conductance sensor and the continuous wave ultrasonic Doppler (CWUD) sensor is proposed. In this method, the signals of water holdup and Doppler frequency shift are respectively obtained by conductance sensor and CWUD sensor. In order to extract rich features of flow regimes, empirical wavelet transform (EWT) and Fourier spectrum analysis are used to process the signals. The statistical method is used to get the feature vectors, and the similarity of feature vectors is compared by Mahalanobis distance to obtain the identification rate of the test samples of bubbly flow, plug flow, and slug flow. Compared with only using the conductance sensor, this method used two types of sensors can obtain more comprehensive flow characteristics, and have a higher flow regimes identification rate.