Fast FMCW Terahertz Imaging for In-Process Defect Detection in Press Sleeves for the Paper Industry and Image Evaluation with a Machine Learning Approach

Abstract

We present a rotational terahertz imaging system for inline nondestructive testing (NDT) of press sleeves for the paper industry during fabrication. Press sleeves often consist of polyurethane (PU) which is deposited by rotational molding on metal barrels and its outer surface mechanically processed in several milling steps afterwards. Due to a stabilizing polyester fiber mesh inlay, small defects can form on the sleeve’s backside already during the initial molding, however, they cannot be visually inspected until the whole production processes is completed. We have developed a fast-scanning frequenc-modulated continuous wave (FMCW) terahertz imaging system, which can be integrated into the manufacturing process to yield high resolution images of the press sleeves and therefore can help to visualize hidden structural defects at an early stage of fabrication. This can save valuable time and resources during the production process. Our terahertz system can record images at 0.3 and 0.5 THz and we achieve data acquisition rates of at least 20 kHz, exploiting the fast rotational speed of the barrels during production to yield sub-millimeter image resolution. The potential of automated defect recognition by a simple machine learning approach for anomaly detection is also demonstrated and discussed.

A Priori Information and Off-Axis Measurements in Terahertz Tomography

“…The inclusion of the above-mentioned a priori information leads to an improved model of the forward problem of tomography represented by (7). The first approach, method 1, is similar to what has been done in [21]: When building the matrix 𝐴 ̅ from the trajectories simulated by our Ray tracing algorithm (Algorithm 2), we incorporate the deviated beam paths by connecting the measurement value in the sinogram with the pixels the deviated beam passes.…”

“…More exhaustive coverage of the derivations above can be found in [18], [22]- [24]. The forward problem of terahertz tomography can thus be described by a simple matrix multiplication 𝐴 ̅ 𝑥 ⃗ = 𝑏 ⃗⃗ (7) Herein, the vector 𝑏 ⃗⃗ denotes the sinogram containing the measurement data, the unknown vector 𝑥 ⃗ represents the image to be reconstructed and the matrix 𝐴 ̅ links the corresponding entries of 𝑏 ⃗⃗ and 𝑥 ⃗. The latter contains the trajectories of the rays describing how the terahertz rays travel through the measurement area.…”

“…The CGLS algorithm is an iterative algorithm for approximating the solution 𝑥 𝑖 of an inverse problem such as (7), for a known positive-definite (not necessarily square) matrix 𝐴 ̅ and an also known measurement vector 𝑏 ⃗⃗ . It does so by updating an initially given solution 𝑥 0 going in steps, which are all 𝐴 ̅ 𝑇 𝐴 ̅ -conjugate to each other.…”

Object Recognition in High-Resolution Indoor THz SAR Mapped Environment

“…Complementary to the indoor THz SAR applications, many novel THz SAR testbeds are proposed in the literature in areas such as the automobile [ 11 ], non-destructive testing [ 12 ], and security [ 13 ]. The object recognition serving security purposes is presented in [ 14 , 15 , 16 , 17 ].…”

“…The prime focus in [ 14 , 15 , 16 , 17 ] is the recognition of objects beneath the clothes targeting security applications. Recently, object analysis at the THz spectrum from the NDT perspective is presented in [ 12 ]. In the automotive field, object recognition based on radar-cross-section (RCS) sensing is of significant interest and powered by the commercially available radar chips at 77 GHz [ 18 ].…”