Research On The Terahertz Imaging System And The Image Super-resolution Reconstruction Algorithm

Terahertz(THz)wave has many advantages such as strong penetration,low photon energy and so forth.THz imaging has shown great application prospects in Non-Destructive Testing(NDT),safety inspection and other fields.However,the existing terahertz imaging systems have their own shortcomings,such as slow imaging speed,low resolution,and more complex hardware.At present,the scanning imaging is often used in terahertz security,which is difficult to be real-time and the imaging time of which is longer.The complex structure and high cost of the terahertz detector is one of the factors that the terahertz system developers must take into account.Many scholars have focused on how to improve the resolution of terahertz imaging.Traditionally,the resolution of the image can be improved from two aspects of the system and the algorithm.From the view of hardware,this paper effectively improves the speed and resolution of the system by designing the terahertz imaging system.However,the wavelength of terahertz wave is longer.Even if the imaging system reaches the diffraction limit,the imaging quality of the system is still poor.Therefore,it is necessary to reconstruct the degraded terahertz image with super resolution reconstruction algorithm from the view of software.It will improve the resolution of terahertz image without increasing the complexity and cost of the system.This article carries out the work mainly from the two aspects of the system and the super-resolution reconstruction algorithm.The relevant aspects of terahertz imaging system were analyzed and we designed three terahertz optical imaging systems,which are large relative aperture optical transmission system,off-axis parabolic mirror reflection imaging system and the off-axis three-mirror system respectively.At the same time,the terahertz experiment has also been carried out to promote the understanding of the characteristics of the terahertz imaging.In terms of algorithm,the super-resolution reconstruction algorithm based on neural network learning is introduced into the terahertz image reconstruction,and the improved Lucy-Richardson algorithm is used to reconstruct degraded images.