Study Of Terahertz Imaging Of Scrolled/Folded And Laminated Objects

Terahertz tomography has broad application prospects in the field of industry,security,heritage conservation,biomedicine,etc.It has attracted much attention in recent years and become one of the research hotspots of terahertz technology.However,terahertz tomography still faces many challenges for the detection of complex objects,such as scrolled/folded and layered structures.For example,it is difficult to perform nondestructive inspection of scrolled/folded objects;Objects with a layered structure have a problem when imaging,that is the information in the upper layer affects or occludes the information in the lower layer.Around terahertz tomography,this paper carries out research along two main lines: firstly,the non-destructive inspection of scrolled/folded objects based on self-built terahertz computed tomography(THz-CT)imaging system.A solution to the problem of metal artifacts in imaging is proposed based on image segmentation and restoration,according to which a virtual unrolling/unfolding technique for non-destructive detection of scrolled /folded objects is proposed;secondly,reflection imaging research is conducted for layered objects based on self-built terahertz frequency modulated continuous wave(THz-FMCW)imaging system.The FMCW signal processing method is improved by using methods such as multiple autocorrelation detection.Based on this,a multilayer image restoration technique for layered objects is proposed.The main and innovative results achieved are as follows:(1)The advantages of the filtered back-projection algorithm(FBP)are determined by numerical simulation of terahertz computed tomography.A 300 G THz-CT imaging system is independently built to scan four different objects,and the causes of metal artifact are analyzed.Based on this,an effective terahertz metal artifact suppression scheme is proposed.The scheme is based on the image processing perspective and does not require any changes to the system hardware.The algorithm involves a total of eight steps from the acquisition of the original sinogram data to the suppression of metal artifact,the core of which is image segmentation and sinogram restoration.The algorithm is verified by experimentally measuring several metal stubs,using the obtained slice images,and the results shows that the metal artifact is greatly suppressed.(2)To solve the problem that scrolled/folded objects are difficult to be inspected nondestructively,this paper abstracts the object as a curve extrude structure model,whereby a virtual unrolling/unfolding technique algorithm based on THz-CT imaging is designed,including slice image reconstruction,metal artifact suppression,ridgeline detection,and image stitching in a total of four stages.Experimental measurements of a self-made oil painting scroll object as well as a folded paper object are carried out using an independently built 300 G THz-CT imaging system to verify the feasibility of the algorithm,and both generated correct virtual unrolled/unfolded images,realizing a terahertz non-destructive detection technique for scrolled/folded objects.This technique explores a new application scenario for THz-CT imaging technology and provides an alternative new solution for the related needs in the field of heritage conservation or security.(3)A THz-FMCW imaging system with a sweep range of 120～160G is built;a set of THz-FMCW signal processing method based on multiple autocorrelation detection is proposed,including pre-processing,spectral analysis and imaging.It is also verified by metal plate translation experiment and focal plane imaging experiment,showing that the proposed method can effectively suppress system noise.(4)The problem that the information of the front layer affects or occludes the information of the back layer in the THz-FMCW reflection imaging process of layered objects is investigated,and an image restoration-based occlusion information removal method is proposed.A self-made double-layered object is constructed,and experimental measurements are carried out with no/partial occlusion of information using an independently built THz-FMCW imaging system,and the images are recovered using the proposed occlusion information removal method.The results show that the recovered image can effectively remove the influence of the information in the front layer in the case of no information coverage;in the case of partial coverage,the recovered image can restore the information in the occluded layer.This scheme can be extended to multi-layer image stacks and is expected to solve the prevalent problem of information occlusion in reflective imaging.