Differentiation of white architectural paints by microscopic laser Raman spectroscopy and chemometrics

Non-destructive detection of single corn seed vigor based on visible/near-infrared spatially resolved spectroscopy combined with chemometrics

Seed vigor is an essential quality evaluation index for seed selection. However, accurately detecting the vigor of a single corn seed is challenging. In this study, we constructed a single-fiber spatially resolved detection device using visible/near-infrared spectroscopy to investigate the patterns and correlations between spatially resolved spectroscopy (SRS) at 500-1000 nm and seed vigor. The device collected spectral data at a light source-detector distance of 5-6.6 mm on the embryo side (S1) and endosperm side (S2) of the corn seeds. The proposed spectral ratio method based on SRS and spectral combination analysis achieved an improvement in the detection accuracy of different corn seed vigor. Modeling by SG-CARS-PLSDA using the ratio method showed further improvement in the prediction ability. The highest accuracy for both S1 and S2 in the Zhengdan 958 variety was 91.67 %, while those of S1 and S2 for the Shaandan 650 variety were 86.67 % and 88.33 %, respectively. In addition, SRS was found to be more advantageous in S2 acquisition, verifying the potential of SRS in the non-destructive testing of seed vigor. This provides a favorable reference for the comprehensive evaluation of other internal quality indices of seeds.

Bayesian Fusion Model Enhanced Codfish Classification Using Near Infrared and Raman Spectrum

In this study, a Bayesian-based decision fusion technique was developed for the first time to quickly and non-destructively identify codfish using near infrared (NIRS) and Raman spectroscopy (RS). NIRS and RS spectra from 320 codfish samples were collected, and separate partial least squares discriminant analysis (PLS-DA) models were developed to establish the relationship between the raw data and cod identity for each spectral technique. Three decision fusion methods: decision fusion, data layer or feature layer, were tested and compared. The decision fusion model based on the Bayesian algorithm (NIRS-RS-B) was developed on the optimal discrimination features of NIRS and RS data (NIRS-RS) extracted by the PLS-DA method whereas the other fusion models followed conventional, non-Bayesian approaches. The Bayesian model showed enhanced classification metrics (92% sensitivity, 98% specificity, 98% accuracy) that were significantly superior to those demonstrated by any of other two spectroscopic methods (NIRS, RS) and the two data fusion methods (data layer fused, NIRS-RS-D, or feature layer fused, NIRS-RS-F). This novel proposed approach can provide an alternative classification for codfish and potentially other food speciation cases.

A novel approach for forensic identification of automotive paints using optical coherence tomography and multivariate statistical methods

Automotive paint is one of the most important evidence in solving vehicle-related criminal cases. It contains the critical information about the suspected vehicle, providing essential clues for the investigation. In this study, a novel approach based on optical coherence tomography combined with multivariate statistical methods was proposed to facilitate rapid, accurate and nondestructive identification of different brands of automotive paints. 164 automotive paint samples from 8 different manufacturers were analyzed by a spectral-domain optical coherence tomography system (SD-OCT). Two-dimensional cross-sectional OCT images and three-dimensional OCT reconstruction of vehicle paints of different paints were obtained to show the internal structural differences. Visual discrimination of A-scan data after registration and averaging processing was first used to distinguish different samples. An scanning electron microscope was utilized to obtain the cross-sectional image of the sample to evaluate the effectiveness of OCT technique. Then the original A-scan data, first derivative data and second derivative data of 136 paints with four layers from 7 different manufacturers were collected. Multivariate statistical methods, including principal component analysis (PCA), multi-layer perceptron (MLP), k-nearest neighbor (KNN) algorithm and Bayes discriminant analysis (BDA), were used to analyze different datasets. The results show the hybrid PCA and BDA model based on the first derivative OCT data achieved the best result of 100% accuracy on the testing dataset for identifying automotive paints. It is demonstrated that the OCT technique combined with multivariate statistics could be a promising method for identifying the automotive paints rapidly and accurately.