Improving Time Resolution of Terahertz Reflected Signal based on Sparse Reconstruction

Terahertz sparse reconstruction method (SR) of impulse response sequence based on sparse representation is presented to locate and separate the complex overlapping echoes in the fault detection in multilayer structures. Considering the convolution model and sparse prior knowledge of impulse response sequence, this reconstruction problem of impulse response sequence can be regarded as a convex problem, which is solved by L1-norm minimization. Numerical simulations and experimental measurements validate the effectiveness and applicability of SR method and show that SR method provide a power tool for terahertz nondestructive testing of multilayered composite structures.