Depth Data Denoising in Optical Laser Based Sensors for Metal Sheet Flatness Measurement: A Deep Learning Approach.

Surface flatness assessment is necessary for quality control of metal sheets manufactured from steel coils by roll leveling and cutting. Mechanical-contact-based flatness sensors are being replaced by modern laser-based optical sensors that deliver accurate and dense reconstruction of metal sheet surfaces for flatness index computation. However, the surface range images captured by these optical sensors are corrupted by very specific kinds of noise due to vibrations caused by mechanical processes like degreasing, cleaning, polishing, shearing, and transporting roll systems. Therefore, high-quality flatness optical measurement systems strongly depend on the quality of image denoising methods applied to extract the true surface height image. This paper presents a deep learning architecture for removing these specific kinds of noise from the range images obtained by a laser based range sensor installed in a rolling and shearing line, in order to allow accurate flatness measurements from the clean range images. The proposed convolutional blind residual denoising network (CBRDNet) is composed of a noise estimation module and a noise removal module implemented by specific adaptation of semantic convolutional neural networks. The CBRDNet is validated on both synthetic and real noisy range image data that exhibit the most critical kinds of noise that arise throughout the metal sheet production process. Real data were obtained from a single laser line triangulation flatness sensor installed in a roll leveling and cut to length line. Computational experiments over both synthetic and real datasets clearly demonstrate that CBRDNet achieves superior performance in comparison to traditional 1D and 2D filtering methods, and state-of-the-art CNN-based denoising techniques. The experimental validation results show a reduction in error than can be up to 15% relative to solutions based on traditional 1D and 2D filtering methods and between 10% and 3% relative to the other deep learning denoising architectures recently reported in the literature.

Optical Dual Laser Based Sensor Denoising for OnlineMetal Sheet Flatness Measurement Using Hermite Interpolation.

Flatness sensors are required for quality control of metal sheets obtained from steel coils by roller leveling and cutting systems. This article presents an innovative system for real-time robust surface estimation of flattened metal sheets composed of two line lasers and a conventional 2D camera. Laser plane triangulation is used for surface height retrieval along virtual surface fibers. The dual laser allows instantaneous robust and quick estimation of the fiber height derivatives. Hermite cubic interpolation along the fibers allows real-time surface estimation and high frequency noise removal. Noise sources are the vibrations induced in the sheet by its movements during the process and some mechanical events, such as cutting into separate pieces. The system is validated on synthetic surfaces that simulate the most critical noise sources and on real data obtained from the installation of the sensor in an actual steel mill. In the comparison with conventional filtering methods, we achieve at least a 41% of improvement in the accuracy of the surface reconstruction.

Erratum: Topological Optical Waveguiding in Silicon and the Transition between Topological and Trivial Defect States [Phys. Rev. Lett. 116, 163901 (2016)].

This corrects the article DOI: 10.1103/PhysRevLett.116.163901.

Topological Optical Waveguiding in Silicon and the Transition between Topological and Trivial Defect States.

One-dimensional models with topological band structures represent a simple and versatile platform to demonstrate novel topological concepts. Here we experimentally study topologically protected states in silicon at the interface between two dimer chains with different Zak phases. Furthermore, we propose and demonstrate that, in a system where topological and trivial defect modes coexist, we can probe them independently. Tuning the configuration of the interface, we observe the transition between a single topological defect and a compound trivial defect state. These results provide a new paradigm for topologically protected waveguiding in a complementary metal-oxide-semiconductor compatible platform and highlight the novel concept of isolating topological and trivial defect modes in the same system that can have important implications in topological physics.

The finite element method applied to the study of two-dimensional photonic crystals and resonant cavities.

A critical assessment of the finite element (FE) method for studying two-dimensional dielectric photonic crystals is made. Photonic band structures, transmission coefficients, and quality factors of various two-dimensional, periodic and aperiodic, dielectric photonic crystals are calculated by using the FE (real-space) method and the plane wave expansion or the finite difference time domain (FDTD) methods and a comparison is established between those results. It is found that, contrarily to popular belief, the FE method (FEM) not only reproduces extremely well the results obtained with the standard plane wave method with regards to the eigenvalue analysis (photonic band structure and density of states calculations) but it also allows to study very easily the time-harmonic propagation of electromagnetic fields in finite clusters of arbitrary complexity and, thus, to calculate their transmission coefficients in a simple way. Moreover, the advantages of using this real space method in the context of point defect cluster quality factor calculations are also stressed by comparing the results obtained with this method with those obtained with the FDTD one. As a result of this study, FEM comes out as an stable, robust, rigorous, and reliable tool to study light propagation and confinement in both periodic and aperiodic dielectric photonic crystals and clusters.

Phase-shifted imaging on multi-directional induction thermography.

A novel multi-directional eddy current thermography (ECT) system is presented generating sets of directional phase images that have been fused with a processing pipeline allowing for an improved probability of detection (POD). Inhomogeneous electromagnetic Joule heating derived from the diversion of induced eddy currents provoked by cracks, altering its path around as well as under its bottom, is the principal phenomenon enabling its usage as a non-destructive-evaluation (NDE) technique. Most induction thermography systems employ inductors derived from old designs, optimized for localized heating with a fixed magnetic field direction. This provokes a directional detection blind-spot for surfaces with random crack orientations. In this paper we have observed that the pattern associated with the thermal response distribution can be geometrically correlated to the relative orientation of the magnetic field regarding the crack, conforming to a rotating feature that has not been described before. Extracting the apparent motion as an optical flow, with a phase-shifting interpolation of the intermediate orientations, generates a signal that enables a robust segmentation of a wide variety of defects in ferritic and austenitic alloys. Its performance has been evaluated with two 'Hit/Miss' POD studies TIG welds Inconel 718 and Haynes 282 alloys. Results show an increased detectability regarding the manual labelling of the defects in the same directional set, employing the same input.