基于改进双边滤波的GPR图像强振幅噪声压制算法  

GPR image strong amplitude noise suppression algorithm based on improved bilateral filtering

作　　者：刘毛毛[1] 刘宗辉[1] 周东[1] 刘保东 黎永健 LIU MaoMao;LIU ZongHui;ZHOU Dong;LIU BaoDong;LI YongJian(College of Civil Engineering and Architecture,Guangxi University,Nanning 530004,China;Nanning Pipe Gallery Construction Investment Co.,Ltd.,Nanning 530219,China;Pavement Branch of Guangxi Road and Bridge Engineering Co.,Ltd.,Nanning 530200,China)

机构地区：[1]广西大学土木建筑工程学院,南宁530004 [2]南宁城建管廊建设投资有限公司,南宁530219 [3]广西路桥工程集团有限公司路面分公司,南宁530200

出　　处：《地球物理学进展》2022年第4期1749-1756,共8页Progress in Geophysics

基　　金：国家自然科学基金项目(51708136);广西科技基地和人才专项(桂科AD19245153)联合资助。

摘　　要：目前探地雷达数据处理更多地是直接引入其他领域的图像和信号处理方法,如中值滤波、小波变换和剪切变换等去噪方法.存在效率低、图像边缘保持差、噪声处理不彻底等缺点,且滤波效果取决于噪声的类型.为此,提出了一种结合探地雷达倾角信息和相关性信息的改进双边滤波算法.首先,分别基于梯度平方的结构张量算法和相似系数分析方法提取了探地雷达局部倾角信息和相关性信息;然后利用所提取的角度信息构建高斯角度相似权值函数,并将该函数代替双边滤波算法中的高斯空间权值函数,增强噪声压制的方向性;将双边滤波算法中的高斯像素相似度权值函数中心点代替为邻域内的中值,提高雷达图像中的奇异点噪声压制能力;最后在权值函数中结合相关系数信息,进一步压制强振幅噪声.数值模拟和实际数据分析结果表明,本文改进算法能够有效压制随机噪声和强振幅噪声,增强振幅的连续性.Nowadays many image or signal processing methods are directly introduced in noise suppression of Ground Penetrating Radar(GPR)image,such as median filtering,wavelet transform,and shearlet transform.These methods have some defects,for example low efficiency,poor image edge retention,and incomplete noise processing.Besides,the filtering effect mainly depends on the type of noise.Therefore,we propose in this paper an improved bilateral filtering algorithm based on GPR angle information and correlation analysis.First,based on the gradient squared structure tensors algorithm and the method of similarity coefficient analysis respectively,the information of local dip and correlation of GPR data is extracted.Then the extracted angle information is used to construct the weight function of gaussian angle similarity.To enhance the directionality of noise suppression,the weight function of gaussian angle similarity is used to replace the weight function of gaussian space in the bilateral filtering algorithm.Besides,to suppress the singular point noise,the center point of the weight function of gaussian pixel similarity in the bilateral filtering algorithm is replaced with the median value in the neighborhood.Finally,the correlation coefficient matrix is introduced into the weight function to further suppress the strong amplitude noise.The reliability of the proposed method was tested with synthetic and real GPR data.The analysis results show that the improved algorithm in this paper can not only suppress random noise,but also has a better suppression effect on strong amplitude noise.

关 键 词：探地雷达 噪声压制 结构张量 相关性分析 改进双边滤波 

分 类 号：P631[天文地球—地质矿产勘探]