A signal filtering algorithm ( SVMD-EWT) based on successive variational mode decomposition and empirical wavelet transform is proposed to solve the harmonic separation problem of wideband RF echo signal in ultrasonic tissue harmonic imaging. The signal is decomposed by successive variational modes to acquire the center frequencies of narrow-band modes. The modes are classified according to the local minimum found by Scale-space in empirical wavelet transform. The energy of the modes aliasing in fundamental and harmonic components is chosen as the parameter to optimize the mode boundary of the empirical wavelet transform. Then the empirical wavelet filter is designed to filter the ultrasonic RF echo signals. Simulation and experiments show that the proposed method has better filtering performance than the traditional high-pass filter method with artificially given cutoff frequency and the pulse inversion method dealing with paired echo signals generated by transmitting the inverse phase signal. The contrast of the harmonic breast ultrasound images generated after bandpass filtering and the method proposed in this study are 15. 77 dB and 20. 78 dB, respectively.