Detection of early bruises in jujubes based on reflectance, absorbance and Kubelka-Munk spectral data
Introduction
Bruising is defined as damage to fruit tissue due to external forces that cause physical changes in texture or chemical changes in color, smell and taste (Baranowski et al., 2013). Moreover, bruising of fruit was usually caused in during harvest, transportation, cleaning, or on the sorting line by impact, compression, vibration, or abrasion (Brosnan and Sun, 2004; Gowen et al., 2007; Van Zeebroeck et al., 2007; Celik et al., 2021). Such damage means that the fruit are more likely to be invaded by the appearance, water loss and risk of bacterial and fungal contamination (Luo et al., 2012), which will not only cause the affected fruit to rot, but will also affect other intact fruit (Zhang et al., 2017). Meanwhile, damaged tissues need a few hours to become darkened and brownish due to the enzymatic or chemical oxidation of phenolic compounds (Toivonen et al., 2007; Baranowski et al., 2013; Opara and Pathare, 2014). Lingwu long jujube is the dominant characteristic jujube fruit in Ningxia, China, and is important economic, social and scientific research significance (Li et al., 2007; Wu et al., 2016). However, jujubes were prone to damage in harvesting, transportation and other processes (Mo et al., 2017; Li et al., 2018). External damage can be easily picked out, but internal bruising is not easily identified. Consequently, it is necessary to distinguish bruised fruit from undamaged fruit to improve fruit quality and prevent food contamination.
Traditionally, bruised jujubes were identified through visual inspection and removed manually, which was a laborious task. Therefore, there was an urgent demand for machine vision technology for rapid and nondestructive, early bruise detection in jujubes (Keresztes et al., 2016). Machine vision, an advanced technology to “see” objects with the assistance of computers, had been used in agriculture, including preharvest and postharvest product-quality and safety detection and sorting (Rehman et al., 2019). Optical techniques hold great promise for the nondestructive evaluation of bruise susceptibility because they are generally rapid and nondestructive or noninvasive, and more importantly, they could provide a large amount of information about the internal conditions of fruit (Huang and Lu, 2010). In particular, hyperspectral imaging (HSI), also known as chemical or spectroscopic imaging, represents an emerging technique that provides both spatial information of imaging systems and spectral information of spectroscopy (Grahn and Geladi, 2007; Ferrari et al., 2015; Zhang et al., 2019b; Mogollon et al., 2020).
Over the past few years, some researchers have reported bruise detection in fruit using the HIS system (Leiva-Valenzuela et al., 2013; Lee et al., 2014; Hu et al., 2016). For example, HSI was used for online early bruise detection in apples (Keresztes et al., 2016). With the best combination of the first derivative and mean centering, followed by image postprocessing, this system could detect fresh bruises in thirty apples with 98.0 % accuracy at the pixel level with a processing time per apple below 200 ms. Fan (Fan et al., 2017) utilized near-infrared (NIR) hyperspectral reflectance imaging (950−1650 nm) with reduced spectral features to detect internal bruising in blueberries after a mechanical impact of 30 min to 12 h. The overall discrimination accuracy for healthy and bruised blueberries in the validation set was 93.3 % and 98.0 %, respectively. The accuracies of the healthy and bruised band ratio images were 93.3 % and 95.9 %, respectively. By comparison, hyperspectral contains image and spectral data, which provides sufficient information for Lingwu long jujube bruise classification. However, it also includes other background information, noise, and useless information. It was thus necessary to preprocess and reduce the dimension of the original spectrum.
The present work was aimed at comparing three kinds of spectral datasets (R, A, and K-M) for the identification and classification of the grade of bruising in Lingwu long jujube. Specific objectives were to preprocess three kinds of spectral data by different pretreatment algorithms and select characteristic wavelengths, and then establish classification and prediction models of partial least squares-discriminant analysis (PLS-DA) and support vector machine (SVM) based on three kinds of spectra for the degree of impact damage of Lingwu long jujubes, which will provide theoretical guidance for online bruising detection of Lingwu long jujubes.
Section snippets
Jujube samples
Lingwu long jujubes were picked from a local orchard in Ningxia, China. Fresh jujubes were harvested at optimum maturity (the jujubes were all red with three-star jujube, longitudinal diameter range/mm was 36 ∼ 42 mm and transverse diameter range/mm was 23 ∼ 88 mm) and brought to the laboratory in a cooler. Fruits having similar in size and freeing from any mechanical defects and diseases were selected for the experiment. A total of 278 jujubes were used for hyperspectral analysis. All jujubes
Spectral analysis of Lingwu long jujube
As depicted in Fig. 2, the R, A, and K-M spectra curves of sound jujube and various bruise jujube (level I, II, and III bruises) showed the same trends. The intensity values of the reflectance spectra in sound jujubes were lower than those in bruised jujubes, mainly due to the cell membrane rupture. The results were in agreement with the findings in apples (Luo et al., 2012), strawberries (Nanyam et al., 2012), and pears (Lee et al., 2014). The spectral curves overlapped between 600 nm and
Conclusions
This study observed and compared the ability between R, A and K-M of Vis/NIR hyperspectral imaging for early bruise detection of Lingwu long jujubes. Through processing and modeling the obtained spectral data, it was found that the A-Raw-iVISSA-PLS-DA model had the best effect, and the number of feature variables was the least, accounting for 28.8 % of the total wavelength. The model’s calibration set and cross validation set accuracies were 88.9 % and 100.0 %, respectively, and the
Data availability
No data was used for the research described in the article.
CRediT authorship contribution statement
Ruirui Yuan, Mei Guo and Chengyang Li: Roles/Writing-original draft, Writing-review and editing, Experiment; Shoutao Chen, Jianguo He and Guishan Liu: Experimental scheme design; Guoling Wan and Naiyun Fan: Review and editing, Experiment.
Declaration of Competing Interest
The authors declare that we have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Acknowledgements
This work was supported by the Key R&D Program of Ningxia Hui Autonomous Region in 2018, “Research and development of key technology and equipment for cold chain storage of typical fruit and vegetables in Ningxia” (Grant No. 2018BCF01001), and the National natural science foundation of China (No: 3216160344) for funding.
References (54)
- et al.
Supervised classification of bruised apples with respect to the time after bruising on the basis of hyperspectral imaging data
Postharvest Biol. Technol.
(2013) - et al.
1H NMR and multi-technique data fusion as metabolomic tool for the classification of golden rums by multivariate statistical analysis
Food Chem.
(2020) - et al.
Vibrational spectroscopic methods for the overall quality analysis of washing powders
Talanta
(2016) - et al.
Improving quality inspection of food products by computer vision-a review
J. Food Eng.
(2004) - et al.
Effects of bruising of ‘Pink Lady’ apple under impact loading in drop test on firmness, colour and gas exchange of fruit during long term storage
Postharvest Biol. Technol.
(2021) - et al.
Waveband selection for NIR spectroscopy analysis of soil organic matter based on SG smoothing and MWPLS methods
Chemometr. Intell. Lab.
(2011) - et al.
Early detection of apple bruises on different background colors using hyperspectral imaging
LWT - Food Sci. Technol.
(2008) - et al.
Detection of blueberry internal bruising over time using NIR hyperspectral reflectance imaging with optimum wavelengths
Postharvest Biol. Technol.
(2017) - et al.
On the geometry of SNV and MSC
Chemometr. Intell. Lab.
(2009) - et al.
Fast exploration and classification of large hyperspectral image datasets for early bruise detection on apples
Chemometr. Intell. Lab.
(2015)
Hyperspectral imaging - an emerging process analytical tool for food quality and safety control
Trends Food Sci. Technol.
A tutorial review: metabolomics and partial least squares-discriminant analysis-a marriage of convenience or a shotgun wedding
Anal. Chem. Acta.
Classification and characterization of blueberry mechanical damage with time evolution using reflectance, transmittance and interactance imaging spectroscopy
Comput. Electron. Agr.
Apple mealiness detection using hyperspectral scattering technique
Postharvest Biol. Technol.
Quantitative analysis of yeast fermentation process using Raman spectroscopy: comparison of CARS and VCPA for variable selection
Spectrochim. Acta A. Mol. Biomol. Spectrosc.
PLS-DA vs sparse PLS-DA in food traceability. A case study: authentication of avocado samples
Talanta
Real-time pixel based early apple bruise detection using short wave infrared hyperspectral imaging in combination with calibration and glare correction techniques
Food Control
Hyperspectral near-infrared imaging for the detection of physical damages of pear
J. Food Eng.
Prediction of firmness and soluble solids content of blueberries using hyperspectral reflectance imaging
J. Food Eng.
Nutritional composition of five cultivars of Chinese jujube
Food Chem.
Detection of early bruises on peaches (Amygdalus persica L.) using hyperspectral imaging coupled with improved watershed segmentation algorithm
Postharvest Biol. Technol.
Detection of hidden bruise on kiwi fruit using hyperspectral imaging and parallelepiped classification
Procedia Environ. Sci.
Wavelength selection in vis/NIR spectra for detection of bruises on apples by ROC analysis
J. Food Eng.
Multiplicative scatter correction during on-line measurement with near infrared spectroscopy
Biosyst. Eng.
Metabolomic perturbation precedes glycolytic dysfunction and procreates hyperglycemia in a rat model due to bisphenol S exposure
Environ. Toxicol. Phar.
Firmness, dry-matter and soluble-solids assessment of postharvest kiwifruit by NIR spectroscopy
Postharvest Biol. Technol.
On-line fresh-cut lettuce quality measurement system using hyperspectral imaging
Biosyst. Eng.
Cited by (31)
Comparative study of Vis/NIR reflectance and transmittance method for on-line detection of strawberry SSC
2024, Computers and Electronics in AgricultureAn overview on optical non-destructive detection of bruises in fruit: Technology, method, application, challenge and trend
2023, Computers and Electronics in AgricultureIdentification and quantification of adulterated Tieguanyin based on the fluorescence hyperspectral image technique
2023, Journal of Food Composition and AnalysisHighly selective and sensitive determination of ceftriaxone sodium using nitrogen-rich carbon dots based on ratiometric fluorescence
2023, TalantaCitation Excerpt :Therefore, accurate and reliable determination of ceftriaxone sodium in biological and food samples is of great importance. At present, there are several analytical methods for detecting antibiotic residues, including high-performance liquid chromatography (HPLC) [1], capillary electrophoresis [2], microbial determination absorbance [3], chemiluminescence [4], and electrochemical methods. However, most of these methods are time-consuming, expensive, and even require more complex equipment, which poses a great challenge to the realization of rapid and convenient determination.
- 1
These authors contributed equally to this work.