PROCESSING OF DETAIL WAVELET-COEFFICIENTS TO IMPROVE THE ACCURACY OF REFLECTOMETRY MEASUREMENTS

Abstract

In the article a modern data processing method of reflectometry measurement of communication line, based on the application of wavelet transform to reflectograms is claimed. This method is based on a multi-level one-dimensional discrete wavelet-decomposition of the reflectogram to the j level (depth) allowing decomposition of the reflectogram into approximation and detail coefficients, containing information on the useful and noise components of the reflectogram. The noise term of the reflectogram is most clearly revealed in the detail coefficients obtained at the lowest decomposition level (j = 1, 2, 3), and which needs to be applied to the threshold processing with different threshold for each coefficient thus the removal of sufficiently small coefficients, which are considered to be noise, is carried out. After this processing of detail coefficients reconstructed reflectogram, with great accuracy, corresponds to the reflectogram without the noise term, that will significantly reduce the localization error of damage and discontinuity of communication line. Evaluation is carried out by comparing mean-square error of recovered, noisy, and original reflectogram without the noise component, as well as on the basis of visual comparison of these reflectograms.