Feature extraction of instantaneous frequency is the key for fault diagnosis of rotating machines,considering that the vibration signals of bearing are usually potential and very weak for dynamic response.A new STFT-SC (Short Time Fourier Transform and Seam Carving) feature extraction method is thus proposed which can estimate instantaneous frequency of vibration signal of rolling bearings.Seam Carving method,usually applied in the field of image processing,is revised according to the characteristics of the rolling bearing.Instantaneous frequency of the vibration signal is estimated using the STFT,then the first order Seam route is solved using the energy gradients of Seam Carving and dynamic planning method;finally,the first order signal is reconstructed and compared with the simulated signal.Results show that the proposed STFT-SC method is effective and quick for estimation of vibration signal even under interference of strong noise and adjacent order components of signal.