Volume 82, Issue 12 - Themenheft: Optical Characterization of Materials / Thomas Längle

The light emission of mineral samples in the visible wavelength range under ultraviolet (UV) illumination is examined by means of hyperspectral imaging. The general physical phenomenon causing this effect is called luminescence. The short-lived kind of luminescence is called fluorescence. Hyperspectral images are images consisting of up to several hundred spectral channels instead of just the three colors in case of common color images. Recently, we presented the first object classification by means of hyperspectral fluorescence analysis of minerals. In this article, previous analyses conducted by us are extended by investigating an increased number of spectral channels and samples of more species. We will present spectral and spatial characteristics of the mineral fluorescence and describe how the identification can possibly be improved.

This work investigates the ability of LIBS to produce quality spectra from small particles of concrete demolition waste using single-shot spectra collected in open air. The 2–8 mm materials are rounded river gravel, green glass shards, and plastic flakes. Considered are focal length, air, moisture, laser energy, and laser incidence angle (LIA) . The research methodology is an experimental study using the so-called observation depth (OD) and spectral abundance as quality indicators. The relation between ablation volume, breakdown threshold, optical signal strength, and OD is captured in a simplified model to provide a better understanding of the dependence of the spectra on the LIA and material positioning in the laser beam. A 100 mm lens provided a compromise between spectral abundance, level of air interference and achievable OD. The study indicates LIBS can yield good quality data, even in cases of up to 3 mm surface roughness. Surface moisture did raise the percentage of bad spectra from an average 4% to 18%, but overall LIBS is still capable of providing quality data under challenging conditions.

The recovery of materials from DW is an important target of the recycling industry and it is important to know which materials are present in order to set up efficient sorting and/or quality control actions. The implementation of an automatic recognition system of recovered products from E nd- O f- L ife (EOL) concrete materials can be a useful way to maximize DW conversion into secondary raw materials. In this paper a new approach, based on H yper S pectral I maging (HSI) sensors, is investigated in order to develop suitable and low cost strategies finalized to the preliminary detection and characterization of materials constituting Demolition Waste (DW) flow stream. Results showed that it is possible to recognize DW materials and to distinguish the recycled aggregates from the investigated contaminants (brick, gypsum, plastic, wood and foam).

Ensuring food safety in various steps along the entire food value chain is crucial to prevent undesired and harmful substances entering the food humans consume. Pesticides and antifungal agents used during growth, processing or along the logistic chain from the field to the consumer can be toxic causing a range of symptoms from stomach pain to the death of the consumer even at trace levels of concentrations. To prevent dangerous contaminants from entering the food chain governmental restrictions on a large number of components have been established and are tightly monitored. For a large number of tests complex and sophisticated equipment is required along with time consuming sample preparation steps, not permitting instantaneous sampling of the specimen at the point of measurement or in a timely manner. High costs and poor reproducibility of commercially available substrates have so far limited the successful application of surface enhanced Raman scattering (SERS) measurements along the food value chain. The use of an affordable handheld Raman instrument with affordable, mass-producible and reproducible SERS substrates will be described in the frame of the contribution with respect to requirements imposed at various stages along the food value chain.

Reducing heterogeneity in harvest material would be beneficial for wine quality and this goal may be achieved through advanced berry sorting systems. The general aim was to assess if a relationship could be found between sugar concentration and hyperspectral images. Grapes were picked at different stages of maturity and the berries were sorted according to their size and density. Hyperspectral images of the berry subsamples were obtained in the vis/NIR wavelength range with a complete spectrum from 400 nm to 2500 nm. Our results showed that vis/NIR can be used to improve the segregation of berries from all tested grape varieties based on their sugar concentration. All berries from all 12 grape varieties were used to train the regression model and the predictive power was tested on all each grape variety separately, while later validated on each variety separately, proving the possibility of using a general regression model with constant parameters to predict sugar concentration. Finally, the impact on quality was tested for red wines. Pinot noir berries with higher sugar concentrations presented more color since anthocyanin concentration was higher. Nevertheless, tannin concentration in skins and seeds tended to decrease. Berries with higher sugar concentration resulted in wines with higher anthocyanin and lower tannin concentration.

Three factors, including tenderness, juiciness and flavour, are found to have an impact on lamb eating quality, which determines the repurchase behaviour of customers. In addition to these factors, the surface colour of lamb can also influence the purchase decision of consumers. From a long time ago, meat industries have been looking for fast and non-invasive objective quality evaluation approaches, where near-infrared spectroscopy (NIRS) and hyperspectral imaging (HSI) have shown great promises in assessing beef quality compared with conventional methods. However, rare research has been conducted for lamb samples. Therefore, in this paper the feasibility of the HSI system for evaluating lamb quality was tested. In total 80 lamb samples were imaged using a visible range HSI system and the spectral profiles were used for predicting lamb quality related traits. For some traits, noise was further removed from HSI spectra by singular spectrum analysis (SSA) for better performance. Considering support vector machine (SVM) is sensitive to high dimensional data, principal component analysis (PCA) was applied to reduce the dimensionality of HSI spectra before feeding into SVM for constructing prediction equations. The prediction results suggest that HSI is promising in predicting some lamb eating quality traits, which could be beneficial for lamb industries.

Near infrared (NIR) spectroscopy has become an analytical tool for material and process control, leading to substantial quality improvements in the output material. In the field of polymer processing NIR spectroscopy has been increasingly applied for on-/in-line monitoring, mainly for lab-scale process development but also for the production of high-value materials. For this paper the bio-polymer polylactide (PLA) was investigated by NIRS in the range of 1.2 to 2.4 μm, in order to identify modifications induced by additives of nanofil ® and nano titanium oxide. Although these additives do not strongly absorb NIR radiation, their influence can be analysed using chemometric codes, due to their light scattering. A twin screw extruder (Haake Polylab PTW 16) was used to extrude polymer pellets mixed with the additives. The extruder was fed by a single screw metering feeder. The optical sensor probes (glass fibers) were placed between the screw and the nozzle to achieve a good optical throughput and to withstand typical temperature and pressure conditions during the extrusion (up to 220 ℃ and a few 10 MPa). The on-line NIR spectroscopy provided real time information which could be used for continuous process or feeding control. The chemometric data evaluation was able to quantify the measured values, especially the content of the additives in the compounds. The established statistical models predicted the actual values with high correlation coefficients (> 0.99). The additives altered the viscosity of the melt and the mechanical properties (Young's modulus) of injection moulded test samples of the compound.

Hyperspectral sensors are becoming cheaper, faster and more readily available. Apart from industry applications, manufacturers push to bring compact devices into the end-consumer market. This development gives rise to many interesting applications such as the identification of counterfeit pharmaceutical products or the classification of food stuffs. These applications require precise models of the underlying classes. However, building these models from expert knowledge is not feasible. In this paper, we propose to use machine learning techniques to infer a model of many classes from an annotated dataset instead. We investigate the use of three popular methods: support vector machines, random forest classifiers and partial least squares. In contrast to similar approaches using support vector machines, we restrict ourselves to the linear formulation and train the classifiers by solving the primal, instead of dual optimization problem. Our experiments on a large dataset show that the support vector machine approach is superior to random forests and partial least squares in classification accuracy as well as training time.