Classification of different tomato seed cultivars by multispectral visible-near infrared spectroscopy and chemometrics

Abstract

The feasibility of rapid and non-destructive classification of five different tomato seed cultivars was investigated by using visible and short-wave near infrared (Vis-NIR) spectra combined with chemometric approaches. Vis-NIR spectra containing 19 different wavelengths ranging from 375 nm to 970 nm were extracted from multispectral images of tomato seeds. Principal component analysis (PCA) was used for data exploration, while partial least squares discriminant analysis (PLS-DA) and support vector machine discriminant analysis (SVM-DA) were used to classify the five different tomato cultivars. The results showed very good classification accuracy for two independent test sets ranging from 94% to 100% for all tomato cultivars irrespective of chemometric methods. The overall classification error rates were 3.2% and 0.4% for the PLS-DA and SVM-DA calibration models, respectively. The results indicate that Vis-NIR spectra have the potential to be used for non-destructive discrimination of tomato seed cultivars with an opportunity to integrate them into plant genetic resource management, plant variety protection or registration programmes.

Detection of live larvae in cocoons of Bathyplectes curculionis (Hymenoptera: Ichneumonidae) using visible/near‐infrared multispectral imaging

“…The performance of each of the three CDA models was evaluated using the classification error rate and accuracy as described in Shrestha et al …”

“…The performance of each of the three CDA models was evaluated using the classification error rate and accuracy as described in Shrestha et al 32 Classification error rate (ER), ER = 1 − Sn+Sp 2 × 100 % where Sn is sensitivity and Sp is specificity Figure 2 shows the fresh weight distribution of the cocoons. As expected, most of the cocoons with dead larvae were lighter and cocoons with live larvae were heavier.…”

Multispectral Imaging for Plant Food Quality Analysis and Visualization

“…For identification of tomato varieties regarding genetic purity, the stepwise PLSDA model displayed an overall classification accuracy of more than 86% (Shrestha and others ). Based on 19 wavelengths ranging from 375 nm to 970 nm, PLSDA and SVM were used to classify 5 different tomato seed cultivars, with a high classification accuracy (over 94%) for all tomato cultivars (Shrestha and others ). Then, Su and Sun () and Erkinbaev and others () identified several feature wavelengths that were finally used as independent variables in PLSDA for segregating organic potatoes from nonorganic tubers and gluten‐free oat kernels from other gluten‐rich grains.…”

Characterisation of Castor (Ricinus communis L.) Seed Quality Using Fourier Transform Near-Infrared Spectroscopy in Combination with Multivariate Data Analysis

“…The model was cross-validated by venetian blinds of 10 data splits with 10 samples in each split. The classification performances of the PLS-DA model were evaluated using sensitivity (Sn), specificity (Sp), classification error rate (CER), classification accuracy as described in Shrestha et al [22] (2016) and the Matthews correlation coefficient (MCC) [23].…”