中文摘要：

针对某大型国有企业的零件在线无损检测工程实际需求,为了提高超声波检测的准确度和精度,引入小波迹理论和匹配追踪算法对超声波检测信号进行消噪处理.首先,在信号的小波变换基础上构建一个小波迹字典;然后,在小波迹域内进行阈值消噪去除信号中的噪声;最后,利用匹配追踪算法通过有限步骤的迭代后,在小波迹字典上用一定数量的小波迹的组合来实现原信号的稀疏描述.小波迹字典内的小波迹具有对信号结构特征无损的描述能力,在小波迹域内消噪克服了传统小波消噪不考虑各尺度之间小波系数的相关性而只进行简单的系数收缩的缺陷.通过仿真试验将本文采用的方法与传统的小波硬阈值和软阈值消噪技术进行了对比,结果表明该方法的消噪效果要优于传统的小波消噪方法.实际超声波检测信号的处理结果也论证了本文所采用消噪技术的优越性.

英文摘要：

In order to improve the nicety and precision of online ultrasonic testing of parts in one big national factory,Wavelet Footprint and Matching Pursuit were employed to denoise the ultrasonic testing signal.Firstly,a wavelet footprint dictionary was constructed from wavelet basis function.Secondly,a threshold was applied in the footprint domain to remove the noise from the noisy signal.At last,by adopting the matching pursuit algorithm,a sparse representation of the testing signal was achieved with a certain number of wavelet footprints in a finite number of iterations in the footprint dictionary.The wavelet footprints characterize efficiently the singular structures of signal.Denoising based on wavelet footprint can better exploit the dependency of the wavelet coefficients across scales than traditional wavelet based denoising which just simply shrinks the wavelet coefficients.The experimental simulation results showed that this method outperformed traditional hard and soft threshold denoising methods.The employed approach was also applied in denoising of actual ultrasonic testing signal.The performance confirmed the usefulness of this approach.