To determine whether superior japonica rice seeds were adulterated by substitution with cheaper seeds or low-quality composition, a strategy based on Terahertz time-domain spectroscopy (THz-TDS) and multivariate classification models was proposed. The spectral signal of 350 seed samples were collected by THz-TDS. The pre-experiments revealed that fisher discriminant analysis (FDA) showed unsatisfactory recognition results. Extreme Learning machine neural network (ELMNN) and support vector machine (SVM) non-linear models were thus developed as multivariate classification methods with the range of 0.11-1.5THz frequency. ELMNN exhibited better performance parameters than SVM in the classification of adulterated samples. In order to intensify the effectiveness of the classification models, three variable selection methods, bootstrapping soft shrinkage (BOSS), interval random frog (IRF) and variable combination population analysis - iteratively retaining informative variables (VCPA-IRIV), were tested and compared. The accuracy of VCPA-IRIV combined with ELMNN was 100%, and the accuracy of VCPA-IRIV combined with SVM was 94.29%. The accuracy of SVM could be further improved by optimization with the Grey Wolf algorithm. The optimized model could achieve 100% classification accuracy of different seed samples. The optimized multivariate models combined with variable selection methods of this study could effectively detect adulterated rice seeds in the field of agriculture.
Abstract The proper identification of car bumper splinters at hit‐and‐run crime scenes is imperative to forensic investigations, as splinters yield crucial pieces of vehicle information that can lead to subsequent investigation resolution and criminal justice. A method based on attenuated total reflectance Fourier transform infrared spectroscopy (ATR‐FTIR) combined with Fisher discriminant analysis (FDA) and support vector machine (SVM) is reported to classify car bumper splinters. The FDA and SVM models were constructed based on full spectrum, fingerprint spectrum, and characteristic spectrum data from 156 car bumper splinter samples. The characteristic spectrum data were extracted by principal component analysis. The classification results for different types of data were compared, and the classification models were analyzed. In the FDA, the model based on the spectral data of the characteristic spectrum yielded the highest classification accuracy, and the classification accuracy based on 10 brands was 88.5%. For polypropylene type; polypropylene, talcum powder, and calcium carbonate type; and polypropylene and talcum powder type bumper samples, the classification accuracy rate reached 97.4%. The classification results were ideal for the SVM, for 10 brands and 3 types of samples, the classification accuracy of the model constructed based on both full spectrum and characteristic spectrum data reached 100%. The results suggest that the SVM model is superior to the FDA model. The SVM model is also suitable for the classification of high‐dimensional data. ATR‐FTIR combined with the chemometrics methods of FDA and SVM is a rapid, nondestructive, and accurate method for the differentiation of car bumper splinters.
The rapid, nondestructive, and accurate classification of pigments in forensic science is important and indispensable. Here, a method for distinguishing different brands and types of pigments was developed by attenuated total reflectance Fourier transform infrared spectroscopy (ATR-FTIR) with chemometrics. A total of 48 pigment samples were collected, and the corresponding infrared spectra were obtained. Baseline correction, multivariate scatter correction, standard normal variable analysis and Savitzky-Golay smoothing were used to preprocess the infrared spectra. Principal component analysis (PCA), factor analysis (FA), Laplacian eigenmaps (LE) and linear discriminant analysis (LDA) were used to extract the characteristic variables of the spectra for the samples. The results were classified by Bayesian discriminant analysis (BDA) and the K-nearest neighbor (KNN) method. The results show that BDA provided a more efficient and accurate model than KNN and the overall classification accuracy was almost 100.0%. Additionally, the classification model was more accurate after extracting the characteristic variables than with the direct use of BDA or KNN. The classification accuracy of gouache and acrylic pigments was 100.0% based on BDA and characteristic variables. The classification accuracy of the BDA and PCA model was 97.2% for two types of gouache pigments and two brands of Picasso gouache pigments. The results indicate that the combination of ATR-FTIR and BDA with a dimensionality reduction method is a potential tool for the classification of different brands and types of pigments.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)