摘要
针对钢轨缺陷超声相控阵法检测应用中所存在的信噪比不高的问题,选用db4小波和阈值法对噪声图像进行小波变换,得到小波系数矩阵,再通过阈值化对小波系数矩阵做逆变换,抑制噪声对检测效果的不利影响,并重构钢轨缺陷的全聚焦图像。选用含人工缺陷的60钢轨试块开展超声相控阵检测实验,对比去噪前后的PSNR和MSE验证本方法的去噪性能,为钢轨缺陷相控阵C扫描成像提供有效保障。
In view of the problems of low signal-to-noise ratio in the diction of rail using ultrasonic phased array, the method of wavelet threshold value was proposed to eliminate the effect of noise. The noisy image was transformed by the db4 wavelet to obtain wavelet coefficient matrix, and then the inverse transform was performed. Finally, the total focusing image of defect was reconstructed. In this work, the No. 60 rail with artificial defects was used for ultrasonic experiments. Compared with original image and traditional median filter algorithm, the experimental results show that the value of SNR and MSE are greatly improved. This method can provide better ultrasonic signals for C scanning of defects in rails using ultrasonic phased array.
作者
陈选民
文晟
CHEN Xuanmin;WEN Sheng(College of Railway Transportation,Guangzhou Railway Polytechnic,Guangzhou 510430,China;Engineering College,South China Agricultural University,Guangzhou 510642,China)
出处
《铁道科学与工程学报》
CAS
CSCD
北大核心
2018年第6期1567-1571,共5页
Journal of Railway Science and Engineering
基金
广东省自然科学基金资助项目(2015A030310182)
广东省科技计划资助项目(2016A020210100)
关键词
60钢轨
无损检测
超声相控阵
去噪
小波阈值法
60 rail
ultrasonic nondestructive detection
ultrasonic phased array
image denoising
threshold
