Passive Millimeter Wave And Terahertz Imaging Super Resolution Algorithm Research

Nowadays the increasingly severer anti-terrorism situation has put forward strict requirements for modern security inspection systems.All the objects emit radiation autoly in the millimeter-wave / terahertz band.Besides having the ability to detection such radiation,passive millimeter wave / terahertz imaging dectection system is high-security,so this system has become a new generation of modern safety inspection methods that can meet the important requirements.However,due to the limitation of the process level of the components required by the current system,the image formed by the system has problems such as blurring and low resolution.In order to solve the problem of low image resolution of passive millimeter-wave/terahertz imaging detection system,this paper is based on the actual passive millimeter-wave detection imaging system and has conducted in-depth research on super-resolution recovery of single-pass passive millimeter-wave images.Wi focus on the two directions of super-resolution recovery algorithm research: First,based on total variation theory of super-resolution recovery algorithm,and second,based on convolutional neural network theory based super-resolution recovery algorithm.It mainly includes the following aspects:(1)Based on the actual passive millimeter-wave detection imaging system,the degrading process and its influencing factors of the system imaging are studied,and the degraded model of the actual passive millimeter wave/terahertz image is established.Based on these,a variety of methods are used to perform super-resolution restoration of passive millimeter-wave images,and the super-resolution recovery effect of each algorithm on millimeter-wave images is analyzed and compared.Their advantages and disadvantages are obtained to from the results.The algorithm is suitable for passive millimeter wave/terahertz images and lays a theoretical foundation.(2)Research and improvement of millimeter-wave image super-resolution restoration algorithm based on total variation.On the basis of the traditional total variational super-resolution algorithm,the L1 norm regularization term is introduced,and the augmented Lagrangian multiplier method is used to optimize the algorithm to accelerate the optimal results and improve the performance of the super-resolution recovery algorithm.And the real-time nature of the algorithm.(3)Based on the classical super-resolution algorithm based on convolutional neural network,this convolutional neural network is optimized and improved for passive millimeter/terahertz images,making it more suitable for passive millimetersv wave/tetahertz image.The simulation verified that the improved algorithm based on convolutional neural network has a good effect on the recovery of passive millimeter-wave/terahertz images.