Hyperspectral images include information enabling the determination of material abundances. Due to the fact that the acquisition of hyperspectral images is time consuming and the processing of these images is computationally costly, we propose an optical approach using spectral filters to retrieve the material abundances. The application of a spectral filter leads to an intensity image encoding estimates for the abundances of a specific material. The acquisition and processing of hyperspectral images becomes superfluous. However, the determination of spectral filters offers a large degree of freedom. In this work, we focus on methods for designing spectral filters incorporating spectral variability. Particularly, we account for reducing the negative effects of spectral variability on the accuracy of estimates for material abundances. According to experimental evaluations, we conclude that including spectral variability into the calculation of spectral filters leads to more accurate abundance estimates when mixed spectra of the considered material mixtures sufficiently fulfill the linear mixing model.

October 17, 2017

Automatic detection of fused tungsten carbides for the characterization of welding process

Maik Rosenberger, Chen Zhang, Karsten Günther, Jean P. Bergmann, Gunther Notni

Page range: 159-166

Abstract

This paper represents an imaged-based method for the quantitative characterization of the dissolution behaviour of fused tungsten carbide in the welding process. For the metallographic investigations based on microscopic images, a hybrid image segmentation procedure for the detection of hard phases is proposed to quantify their amount and distribution in the matrix material. With the help of this tool, the dependence of hard phase characteristic on the welding parameters was investigated.

November 21, 2017

Durchführbarkeitsstudie zur Inline-Überwachung der Wärmeleitfähigkeit und der strukturellen Qualität von Kunststoffen im Heißextrusionsprozess mittels Infrarot-Thermographie

Peter Meinlschmidt, Jochen Aderhold, Friedrich Schlüter

Page range: 167-176

Abstract

Zusammenfassung In der vorliegenden Arbeit wird über erste Ergebnisse eines Forschungsprojektes zur prozessintegrierten Messung der Wärmeleitfähigkeit von heißextrudierten Dämmstoffen und zur Erkennung möglicher struktureller Fehler mittels eines Wärmebildverfahrens berichtet. In Modellversuchen werden Laserimpulse und Kühlfinger benutzt, um auf einem Förderband transportierte Blöcke von Dämmstoffen positiv oder negativ thermisch anzuregen. Mittels einfacher, aber robuster Algorithmen ist es möglich, die Anregungsstellen automatisch zu erkennen und zu verfolgen und die Entwicklung ihrer infraroten Strahlungsintensitäten als Funktion der Zeit zu messen. Diese ist ein Maß für die relativ komplexe Zeitabhängigkeit der Oberflächentemperatur an den Anregungsstellen. Nach einer wahrscheinlich durch die Semitransparenz des Materials bedingten Anlaufphase folgt sie jedoch wenigstens für eine Zeit lang einer gut verstandenen Proportionalität zu t −0.5 , aus der sich die Wärmeleitfähigkeit bestimmen lässt. Damit sind die Grundvoraussetzungen für eine Inline-Messung der Wärmeleitfähigkeit mit diesem Verfahren gegeben.

November 7, 2017

Non-invasive assessment of the bioburden of minced pork using a hand-held fluorescence device

Christina Grimmler, Tamara Kuhfuß, Matthias Heiden, Heinar Schmidt

Page range: 177-183

Abstract

To speed-up monitoring of the hygienic status of meat, fast and non-invasive measuring techniques are required. In this work, the feasibility was investigated to quantify the bioburden of minced meat with a hand-held fluorescence spectrometer. In two independent trials, 144 (165) minced pork samples from 18 (23) batches were stored at 2 ℃ for up to 8 (7) days. Fluorescence spectra, L * a * b * colour and total viable mesophilic plate counts (TVC) were measured in parallel. TVC varied from 2.7 to 10.4 log(TVC/g) in this period. Correlation of the fluorescence spectra with log(TVC/g) using partial least squares regression yielded cross-validated predictions with R 2 cv = 0.72 (0.83) and a prediction error of RMSECV = 0.97 (0.69) log(TVC/g). Cross-validated limits of detection and quantification were determined as 1.0 (0.9) and 3.4 (3.2) log(TVC/g), respectively. This would allow for a quantification well below the alarm threshold of 5.7–6.7 log(TVC/g). The spectra were strongly correlated with storage time with R 2 cv = 0.81 (0.67), but less strong with L * a * b * colour. Storage time was identified as interference which was reduced in the second trial by modifying the sampling.

February 20, 2018

Evaluation and comparison of NIR multi-product calibration methods for Brix prediction

Michael Kopf, Robin Gruna, Thomas Längle, Jürgen Beyerer

Page range: 184-190

Abstract

Near-infrared (NIR) spectroscopy is a widespread technology for fruit and vegetable quality assessment. New fields of application of this technology, like mobile food analysis with handheld low-cost spectrometers, increase the demand for chemometric calibration models that are able to deal with multiple products and varieties thereof at once (so-called multi-product calibration models ). While there are well studied methods for single-product calibration as partial least squares regression (PLSR), multi-product calibration is still challenging. Conventional approaches that work well for single-product calibration can lead to high errors for multi-product calibration. However, nonlinear methods as local regression and artificial neural networks were found to be suitable E. Micklander, K. Kjeldahl, M. Egebo, and L. Norgaard. Multi-product calibration models of near-infrared spectra of foods. Journal of Near Infrared Spectroscopy , 14:395–402, 2006. L. R. Lopez, T. Behrens, K. Schmidt, A. Stevens, J. A. M. Dematte, and T. Scholten. The spectrum-based learner: A new local approach for modeling soil vis-NIR spectra of complex datasets. Geoderma , 195–196:268–279, 2013. . Preliminary studies in multi-product calibration for quantitative analysis of food with near-infrared spectroscopy showed good results for memory-based learning (MBL) and a classification prediction hierarchy (CPH) M. C. Kopf and R. Gruna. Examination of multiproduct calibration approaches for quantitative analysis of food with near infrared spectroscopy. Bachelor's thesis, Karlsruhe Institute of Technology KIT, 2016. . In this study, three varieties of apples, pears and tomatoes with known sugar content (in ○ Brix) are analysed with NIR hyperspectral imaging spectroscopy in the range from 900 nm to 2400 nm. Predictive performance of a linear PLSR model, two nonlinear models (CPH and MBL) and different pre-processing techniques are tested and evaluated. For error estimation, leave-one-product-out and leave-one-out cross-validation are used.

October 12, 2017

THz imaging for recycling of black plastics

Andries Küter, Stefan Reible, Thomas Geibig, Dirk Nüßler, Nils Pohl

Page range: 191-201

Abstract

This paper presents the development of a novel line-camera device in the lower THz-domain which is capable of measuring minute differences in broadband spectral fingerprints of non-conducting materials. The primary focus is sorting black plastics in industrial recycling contexts, where large scale sorting of different types of black plastics remains a challenge. The system operates between 84 GHz and 96 GHz. As the relevant plastics exhibit no specific absorption lines in this frequency range, a broadband approach is necessary to accumulate slight differences in dielectric properties. Using this technique, enough entropy can be gathered that a machine learning algorithm can be trained to differentiate between different materials. This has to work even in the presence of contaminants such as flame retardants, color pigments and dirt. Preliminary results suggest that the THz sensor system is capable of achieving these goals.

Publicly Available October 5, 2017

Motion-based material characterization in sensor-based sorting

Georg Maier, Florian Pfaff, Florian Becker, Christoph Pieper, Robin Gruna, Benjamin Noack, Harald Kruggel-Emden, Thomas Längle, Uwe D. Hanebeck, Siegmar Wirtz, Viktor Scherer, Jürgen Beyerer

Page range: 202-210

More Download PDF

Abstract

Sensor-based sorting provides state-of-the-art solutions for sorting cohesive, granular materials. Typically, involved sensors, illumination, implementation of data analysis and other components are designed and chosen according to the sorting task at hand. A common property of conventional systems is the utilization of scanning sensors. However, the usage of area-scan cameras has recently been proposed. When observing objects at multiple time points, the corresponding paths can be reconstructed by using multiobject tracking. This in turn allows to accurately estimate the point in time and position at which any object will reach the separation stage of the optical sorter and hence contributes to decreasing the error in physical separation. In this paper, it is proposed to further exploit motion information for the purpose of material characterization. By deriving suitable features from the motion information, we show that high classification performance is obtained for an exemplary classification task. The approach therefore contributes towards decreasing the detection error of sorting systems.

About this journal

Objective

The journal promotes dialogue between the developers of application-oriented sensors, measurement systems, and measurement methods and the manufacturers and measurement technologists who use them.

Topics

The manufacture and characteristics of new sensors for measurement technology in the industrial sector
New measurement methods
Hardware and software based processing and analysis of measurement signals to obtain measurement values
The outcomes of employing new measurement systems and methods