A denoising method of ground penetrating radar b-scan image

Abstract

A ground-penetrating radar B-scan image denoising method, which includes the following steps: Step 1: Ground-penetrating radar GPR detects a single background medium area that is pre-buried with a pipeline target on the surface, and obtains Z A GPR B-scan image constitutes a noise-free GPR label data set; Step 2: utilize each kind of noise-containing GPR data set and its corresponding noise-free GPR label data set to train the multi-scale convolution autoencoder; the multi-scale convolution Autoencoders consist of an encoder E , a decoder D; Step 3: Input the GPR image to be denoised to the trained multi-scale convolutional autoencoder, output the denoised GPR image through encoding and decoding, and calculate the denoised GPR image The signal-to-noise ratio verifies the denoising effect of the multi-scale convolutional autoencoder, and the present invention can effectively denoise the noisy GPR image under the condition of low signal-to-noise ratio.

Description

technical field [0001] The invention belongs to the field of ground-penetrating radar data processing, and specifically discloses a ground-penetrating radar B-scan image denoising method based on a multi-scale convolution self-encoder. Background technique [0002] Ground Penetrating Radar (GPR), as an important technique for underground non-destructive testing, is widely used to study near-surface geophysical structures and detect buried targets; it uses a transmitting antenna to transmit broadband electromagnetic waves to the ground. Scattering occurs when encountering an interface with electromagnetic differences when propagating in the medium. The receiving antenna infers the parameters such as the dielectric properties, spatial position, and structural size of the underground medium and the detection target based on the waveform and delay of the received scattered echo. However, there are often clutter phenomena in the received scattered echo, which seriously affects th...

AI Technical Summary

Problems solved by technology

[0005] In GPR images, commonly used subspace-based denoising methods are singular value decomposition (SVD), principal component analysis (PCA) and independent component analysis (ICA), which utilize various constraints in the cost function Decompose the matrix; the existing research uses the SVD method to decompose the signal, then design the appropriate principal component, and use the PCA method to reconstruct the signal, and realize the PCA-SVD hybrid method for noise reduction, effectively removing the GPR image random noise; recently, Morphological Component Analysis (MCA) was proposed. Morphological Component Analysis has superior performance compared with traditional methods, but it requires prior knowledge for dictionary learning and requires high processing time to remove clutter; therefore it is not suitable for real-time applications

Images