不同小波基对探地雷达数据椒盐噪声处理分析

Comparative Analysis on Ground Penetrating Radar Data Denoising by Different Wavelet Basis within Salt and Pepper Noise

探地雷达因其无损性而广泛应用于工程结构体检测及异常体探测,但其信号易受干扰,常含大量的椒盐噪声,传统时域滤波的去噪效果难以保证,为此将时频域的小波变换应用于探地雷达信号的椒盐噪声去噪处理。通过在正演记录中添加高斯白噪,得到信噪比为1、2、5、10的探地雷达模拟记录。选用不同类型小波基进行滤波处理,并计算滤波后的峰值信噪比,结果显示紧支撑正交小波基去噪效果最好,信噪比为2的模拟记录峰值信噪比值增大至17.3663。进一步将该小波基用于实测记录的去噪处理,实测剖面中90%以上的椒盐噪声被滤除,有效反射得以突出。结果表明:小波变换能应用于探地雷达数据中椒盐噪声的去除,但小波基的选择会对去噪效果产生显著影响,实际应用中值得关注。

Ground penetrating radar(GPR)has been widely used in engineering structure and anomaly body detection because of its nondestructivety.However,the denosing effect of conventional filter in time-domain can not be guaranteed for the susceptible interference of GPR signals by lots of salt and pepper noise.Therefore,the authors used the wavelet transform in time-frequency domain to denoise the salt and pepper noise within GPR data.With the method of adding Gaussian white noise to the forward records,the GPR simulation data were obtained in which the signal-to-noise ratios were 1,2,5 and 10 respectively.Thereafter,the wavelet transform method was used with different wavelet bases to denoise the data and calculate the peak signal-to-noise ratios of the data after denoising.The result shows that the effect of the db10 wavelet basis was best And the peak signal-to-noise ratio of simulation record whose S/N equal to 2 increased to 17.3663 after denosing.When the db10 wavelet basis was further used to denoise the measured records,more than 90 percent salt and pepper noise of the observed profile was denoised and the effective reflection was highlighted.The result shows that the wavelet transform method can be applied to denoise the salt and pepper noise within GPR data.However,the wavelet basis has a distinct influence on the denoise effect,which is worthy of attention in the practical applications.