A blind deconvolution method for attenuative materials based on asymmetrical Gaussian model.

During propagation in attenuative materials, ultrasonic waves are distorted by frequency-dependent acoustic attenuation. As a result, reference signals for blind deconvolution in attenuative materials are asymmetrical and should be accurately estimated by considering attenuation. In this study, an asymmetrical Gaussian model is established to estimate the reference signals from these materials, and a blind deconvolution method based on this model is proposed. Based on the symmetrical Gaussian model, the asymmetrical one is formulated by adding an asymmetrical coefficient. Upon establishing the model, the reference signal for blind deconvolution is determined via maximum likelihood estimation, and the blind deconvolution is implemented with an orthogonal matching pursuit algorithm. To verify the feasibility of the established model, spectra of ultrasonic signals from attenuative polyethylene plates with different thicknesses are measured and estimated. The proposed blind deconvolution method is applied to the A-scan signal and B-scan image from attenuative materials. Results demonstrate that the proposed method is capable of separating overlapping echoes and therefore achieves a high temporal resolution.