Skip to main content
ARS Home » Midwest Area » East Lansing, Michigan » Sugarbeet and Bean Research » Research » Publications at this Location » Publication #343497

Research Project: Nondestructive Quality Assessment and Grading of Fruits and Vegetables

Location: Sugarbeet and Bean Research

Title: Nondestructive measurement of tomato postharvest quality using a multichannel hyperspectral imaging probe

item HUANG, YUPING - Nanjing Agricultural University
item Lu, Renfu
item CHEN, KUNJIE - Nanjing Agricultural University

Submitted to: ASABE Annual International Meeting
Publication Type: Proceedings
Publication Acceptance Date: 7/14/2017
Publication Date: 7/20/2017
Citation: Huang, Y., Lu, R., Chen, K. 2017. Nondestructive measurement of tomato postharvest quality using a multichannel hyperspectral imaging probe. ASABE Annual International Meeting. doi:10.13031/aim.201700195.

Interpretive Summary: Nondestructive measurement of tomato firmness, soluble solids content (SSC) and pH is important, because they directly influence consumer acceptance and the shelf life and processing quality. Visible and near-infrared (Vis/NIR) spectroscopy has been shown to be useful for nondestructively measuring these quality parameters; however, improvements for Vis/NIR measurement are still needed because it only acquires spectral data from a single point or a small area. Recently, a new hyperspectral imaging probe was developed for simultaneous acquisition of 30 spatially-resolved spectra for food samples of flat and/or curve surface over the spectral region of 550-1,650 nm. The probe thus enables better assessment of tissues at different depths and spatial distances. In this research, spatially resolved spectra were acquired, using the new multichannel probe, for 600 'Sun Bright' tomatoes of different maturity stages. Firmness, SSC and pH were measured for the tomatoes using standard destructive methods. Mathematical models for predicting the quality parameters were then developed based on individual spectra and their combinations, from which the optimal single spectrum and two- and three-spectra combinations were determined for firmness, SSC, and pH prediction. Results showed that combinations of two or three spectra resulted in better predictions for all three quality parameters. It was also found that spectra acquired at larger light source-detector distances were more useful for predicting SSC and pH, while firmness prediction was less influenced by the source-detector distance. When using the optimal combinations of spectra, excellent predictions of firmness and pH were obtained with the correlation coefficients of 0.974 and 0.936, respectively. However, lower correlations (up to 0.837) were obtained for SSC prediction, due, in part, to inconsistent measurements by the standard method. This research showed that the new multichannel hyperspectral imaging probe enhanced nondestructive measurement of postharvest quality of tomatoes, and it can also be useful for assessing other horticultural and food products.

Technical Abstract: A multichannel hyperspectral imaging probe with 30 optic fibers covering the wavelength range of 550-1,650 nm and the light source-detector distances of 1.5-36 mm was recently developed for optical property measurement and quality evaluation of food products with flat or curved surface. This paper reports on the application of this multichannel probe for assessing postharvest quality of tomato fruit. Spatially-resolved spectra were acquired for 600 'Sun Bright' tomato fruit of six maturity stages. Firmness, soluble solids content (SSC) and pH of the tomato fruit were measured using standard destructive techniques. Partial least squares models for single spectra and two- and three-spectra combinations were developed and compared to determine the optimal models for prediction of firmness, SSC and pH. Compared with the optimum single spectra models, the optimum combinations of three spectra gave better predictions for firmness and pH, with the correlation coefficient for prediction (r) being improved by 2.2% to 0.974 and by 4.5% to 0.936, respectively, while the standard error of prediction improved by 24.7% and 20.9%, respectively. The optimal three-spectra combination for SSC prediction was relatively poor with r=0.836. SSC and pH predictions were greatly influenced by the light source-detector distance and better predictions were obtained from those channels with large light source-detector distances. The results showed that spatially-resolved hyperspectral imaging probe can be used for measuring quality attributes of tomato fruit.