To solve the problem of the training instability of the Generative Adversarial Network (GAN) in generating ground penetrating radar (GPR) images, the Wasserstein GAN with Gradient Penalty is used to generate the GPR images. Moreover, a new method for constructing the GPR dataset is proposed base on the Finite-difference time-domain method and the measured images. Compared with the original GAN and Wasserstein GAN methods, WGAN-GP has better stability and the generated GPR images are more similar to the actual images. On this basis, the Dense Residual Block (DRB) and the U-Net are combined to propose a Dense Residual Denoising U-Net (DRDU-Net) suitable for GPR images. It uses the coding and decoding process of U-Net to improve the denoising performance. In addition, the introduction of DRB enhances the feature reuse of GPR image and makes U-Net training more stable. The performance of the proposal is evaluated by simulation experiments and compared with the BM3D (Block-matching and 3D) and U-Net. The results show that our proposal has better denoising performance than BM3D and U-Net. When the variance is 20, the peak signal-to-noise ratio increases by about 6.5 dB and 2.4 dB and the structural similarity increases by 0.09 and 0.04, respectively.
在GPR图像去噪领域,学者们常用的方法有传统去噪方法和基于深度学习的去噪方法.其中,常用的传统去噪方法主要有:双边滤波[3]、F-K变换[4]、三维块匹配法(Block-Matching and 3D,BM3D)[5]和基于稀疏表示法[6]等.此类方法能够很好地保留图像的边缘信息,且对噪声的抑制效果较好.但是,也存在很多缺点.比如,双边滤波会移除图像纹理;F-K变换会引起畸变和伪异常现象;三维块匹配算法对非高斯噪声的处理效果不理想,且有较高的计算复杂度;基于稀疏表示法分解出的矩阵存在解释性不强等缺陷.
LEIJ W, FANGH Y, LIY P,et al .GPR forward model of underground non-metallic pipeline based on parallel conformal symplectic Euler algorithm[J].Chinese Journal of Geophysics, 2020, 63(8): 3192-3204.(in Chinese)
LIUM M, LIUZ H, ZHOUD, et al. GPR image strong amplitude noise suppression algorithm based on improved bilateral filtering[J]. Progress in Geophysics,2022,37(4):1749-1756.(in Chinese)
[7]
ZHANGL, LINGT H, YUB,et al .Intensive interferences processing for GPR signal based on the wavelet transform and F-K filtering[J].Journal of Applied Geophysics,2021,186:104273.
[8]
CHENGT, JIAD D .Super-resolution reconstruction based on BM3D and compressed sensing[J].Microscopy,2022,71(5):283-288.
[9]
LIR H, ZHANGH Y, CHENZ,et al .Denoising method of ground-penetrating radar signal based on independent component analysis with multifractal spectrum[J].Measurement,2022,192:110886.
[10]
LUOJ B, LEIW T, HOUF F,et al .GPR B-scan image denoising via multi-scale convolutional autoencoder with data augmentation[J].Electronics,2021,10(11):1269.
[11]
FENGD S, WANGX Y, WANGX,et al .Deep convolutional denoising autoencoders with network structure optimization for the high-fidelity attenuation of random GPR noise[J].Remote Sensing,2021,13(9):1761.
[12]
GURROLA-RAMOSJ, DALMAUO, ALARCÓNT .U-Net based neural network for fringe pattern denoising[J].Optics and Lasers in Engineering,2022,149:106829.
ZENGX H, LIY C, GAOG,et al .Channel adaptive ultrasound image denoising method based on residual encoder-decoder networks[J].Journal of Electronics & Information Technology,2022,44(7):2547-2558.(in Chinese)
[15]
HUANGG, LIUZ, VAN DER MAATENL,et al .Densely connected convolutional networks[C]//2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).Honolulu,HI,USA: IEEE,2017:2261-2269.
[16]
HEK M, ZHANGX Y, RENS Q,et al .Identity mappings in deep residual networks[M]//LEIBE B,MATAS J,SEBE N,et al.Computer Vision-ECCV 2016. Cham:Springer International Publishing,2016:630-645.
[17]
CUIF, CHENY, ZHANGY Q,et al .Research on application of ground penetrating radar array method based on plane beam signal in different geological models[J].Acta Geophysica,2021, 69(6):2241-2260.
[18]
GOODFELLOWI, POUGET-ABADIEJ, MIRZAM,et al .Generative adversarial networks[J].Communications of the ACM,2020,63(11):139-144.
[19]
BROCKNERB, DOWDYJ L, ANDERSOND T,et al .Generative adversarial networks for ground penetrating radar in hand held explosive hazard detection[C]//Detection and Sensing of Mines,Explosive Objects,and Obscured Targets ⅩⅩⅢ.April 15-19,2018.Orlando,USA: SPIE,2018,10628: 306-323.
[20]
GIZZIF T, LEUCCIG .Global research patterns on ground penetrating radar (GPR)[J].Surveys in Geophysics,2018, 39(6):1039-1068.
[21]
QINH, ZHANGD H, TANGY,et al .Automatic recognition of tunnel lining elements from GPR images using deep convolutional networks with data augmentation[J]. Automation in Construction,2021,130:103830.
[22]
WEIX, GONGB Q, LIUZ X, et al. Improving the improved training of wasserstein GANs: a Consistency Term and Its Dual Effect[J].2018.
[23]
ZHANGY L, TIANY P, KONGY,et al .Residual dense network for image super-resolution[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition, June 18-23,2018.Salt Lake City,UT: IEEE,2018:2472-2481.
ZHAOX Q, LIX Y .Image super-resolution reconstruction based on information distillation cascade expansion and compression network[J].Journal of Hunan University (Natural Sciences),2023,50(8):52-61.(in Chinese)