In the elevator wire rope damage inspection, the threshold is the key to distinguish noise and broken wire signals, and the noise reduction effect of the broken wire signal is directly affected by the threshold value. Classical wavelet denoising includes two methods of soft and hard thresholds, but hard threshold processing results have pseudo Gibbs oscillations, soft threshold processing results have fixed deviations. On the basis of existing threshold denoising, soft and hard thresholding methods were improved, and the added characteristic coefficients were added to adjust the size of the wavelet coefficients, a more accurate removal of noise in the signal was achieved. The characteristics of broken wires are retained, and accuracy is provided for the quantitative identification of subsequent wire rope broken wires.