Stationary-wavelet regularized inverse filtering: A robust deconvolution approach for terahertz reflection imaging

A deconvolution process is often applied to extract the impulse function of the target sample in terahertz reflection imaging. Though able to suppress the amplified noise for straightforward inverse filtering, the widely used low-pass or double Gaussian filter suffer from unfavorable over-smoothing and observer-dependent parameter settings. To overcome these, here we propose a new stationary-wavelet regularized inverse filtering approach with effective estimation of the wavelet shrinkage thresholds. Experimental water data validates the good performance of the proposed approach.