According to the analizing of the vibrations in steel strips and its effect on flatness measurment based on the non-contact strip-flatness detection method, a vibration removing method is put forward. The method firstly adopts FIR low-pass filter based on Dolph-Chebyshev Window function designing theory to remove the high frequency noise in the primitive flatness mesurement data. As to consider the influence of random noise, there is additional noise message in low frequency range of flatness mesurement data. Therefore, the method adopts piecewise cubic spline interpolation to revise the data processed by low-pass filter. Application results show that the method have satisfied reliability, can effectively shorten the influence that the strip vibration caused on flatness measurement.