An Improved Back Projection Algorithm for Terahertz Imaging

In this paper, an analytical algorithm based on inverse matrix solution is proposed to perform imaging based on near-field terahertz (THz) scan data. The computational framework of conventional back projection algorithms (BPA) comprises four principal stages. The initial procedure discretising the target region via structured mesh partitioning. This was followed by the subsequent computation of wave propagation time delays for each discrete pixel element. Phase correction operations were then performed to compensate for waveform distortions. Ultimately, image reconstruction was achieved through coherent integral of back propagated signals. In contrast to the conventional BPA, this proposed approach reconstructed the signal propagation model. It achieves this by transforming the back projection process into an inverse problem solution, which is then utilized to solve the forward propagation of the signal. It can solve the inverse problem more accurately by obtaining the actual reflectivity of each pixel of the target. The computed results are closer to the electromagnetic (EM) field distribution of the real scene, reducing the error caused by energy in the sub-aperture signal unrelated to the imaging region compared to BPA. Moreover, the radial resolution has been improved over 30%. Those absorbers, which are generally invisible in radar imaging due to very weak reflections, can be visualized beyond classical BPA. In the imaging process of complex targets, enhanced precision in electromagnetic calculations facilitates more comprehensive reconstructions. The algorithm can be used for elaborate environment imaging and sensing in future.