Super-Resolution Reconstruction For Terahertz Image Based On Gradient Field Transform

In order to improve the quality of terahertz image and overcome its edge blur,this paper studies the super-resolution image reconstruction method,which combines interpolation and gradient field transform,for terahertz image reconstruction.This method reached the purpose of improving the quality of the terahertz images and has a certain practical value.The main research contents and conclusions are as follows:1.The method of super-resolution image reconstruction based on interpolation and gradient field transform is investigated.Considering the drawback of terahertz images,such as low-resolution and edge blur,the method of super-resolution image reconstruction that is based on rational fractal interpolation and gradient field transform is employed for the reconstruction of terahertz images obtained by terahertz transmission imaging system,which expands the application scope of the method.Considering the factors that affect the quality of the gradient field,the spatial entropy-based image enhancement and bilateral filter are introduced to optimize the reconstruction algorithm,which enhance the applicability of the method.Meanwhile,the effects of signal frequency and intensity on the performance of the reconstruction algorithm are analyzed.Experimental results show that this method could recover detail information,sharpen edges and improve quality of terahertz images obtained by terahertz transmission imaging system.Moreover,this method has better performance on improving the quality of images with lower frequency.2.The method of super-resolution reconstruction based on interpolation and gradient field transform is applied on terahertz near-field compressive sensing imaging system.The super-resolution reconstruction method,which combines iterative curvature interpolation and gradient transform,is utilized for the reconstruction of terahertz images obtained by terahertz near-field compressive sensing imaging system and the effect of compressive ratio on the performance of reconstruction is investigated.Experimental results show that this method could improve the quality of images acquired by terahertz near-field compressive sensing imaging system and the performance of the reconstruction would be better when the signal is compressed less.