Abstract: It is theoretically proved that the influence of model boundary the ground penetrating radar(GPR) data acquisition; and based on the theory, designing the physical model within the empty of tunnel lining, probing the model with 3D detection, then the visualization of 3D GPR detection data is realized. We analyze the features of the square hole in the different sections, based on the analysis of CL4 multi-wavelet theory, discuss the necessity of the multi-wavelet preprocessing, and process the CL4 multi-wavelet with balance, using Matlab to realize the CL4 balanced multi-wavelet threshold filtering algorithm; and comparing with the filtering effect based on the conventional Fourier transform. The results show that the 3D detection can determine the size and shape, also including location of the hole, understand the accurate information of hole comprehensive and meticulous, improve the reliability and accuracy of the analyses on radar map, evaluating the harm of hole in lining according to the size and location of defect. The conventional and CL4 multi-wavelet filter can distinguish the unusual target exactly; the target feature has been enhanced with the multi- wavelet transfer, therefore, it results to the cleaner phase axis, more thoroughly eliminate-noise, the same to signal reservation becomes more completely; and the picture also becomes more clearly. The implementation of 3D detection test for the tunnel lining automatic identification technology not only provides a reliable theoretical basis and method, but also has great significance to tunnel detection and quality judicatory appraisal

Key words: lining cavity, 3D detection, model test, CL4 balanced multiwavelet, filtering