Abstract: Ground penetrating radar is applied to detect objectives inside the medium.As there are various interferences and noises in the receipt signal and the reflected wave of the objective tends to be relatively weak so that may be submerged by interferences and noises, it is difficult against data processing and interpretation for ground penetrating radar. And for this reason,eliminating random noise interference turns into a key technology using ground penetrating radar to identify shallow objective.While gives an detailed introduction on the structure and property of 2-D physical wavelet function,this paper also compares the properties of wavelet function which transforms along with the scale,verifies the correctness of the procedures through decomposition and reconstruction of typical signal,provids implementation flow chart concerning threshold de-noising of 2-D physical wavelet frame,carries out de-noising processing through two typical signals,and testifies the effectiveness of this method.