Deformation Monitoring Data De-noising Processing Based on Wavelet Packet

Wavelet packet coefficients carrying real signals have large amplitude but are in minority, while those carrying noise has lower amplitude but is of large number. In this case, the Basic principle of de-noising wavelet packet is to process signals carrying noise. A suitable threshold is chosen in different decomposition level. Wavelet packet coefficient of less than this threshold is set to equal zero, while wavelet packet coefficients of greater than this threshold is reserved and reconstructed into de-noising signals. MSE, SNR, PSNR are regarded as the standards of de-noising evaluation, some mathematical methods such as Shannon entropy, (lp) norm entropy, logarithm energy entropy, threshold entropy, Stein Unbiased Risk Estimate entropy are adopted to measure whether the wavelet packet basis is optimal, minimum Entropy function D value is the best base. Selecting threshold and threshold quantitative is the key to wavelet packet de-noising. And selection of threshold value abides standards such as Sqtwolog, Rigrsure, Heursure, Manimaxi, or Birge-massart. Wavelet packet de-noising method has been applied to tunnel vault sink and landslide monitoring data de-noising processing, which manifests itself being a more elaborate, flexible method compared to wavelet de-noising, since wavelet packet de-noising can even subdivided the low-frequency part and the high-frequency part of upper layer, thus entertains a more precise local analysis capabilities.