Astringency assessment of persimmon by hyperspectral imaging

Abstract

One of the current challenges of persimmon postharvest research is the development of non-destructive methods that allow determination of the internal properties of the fruit, such as maturity, flesh firmness and astringency. This study evaluates the usefulness of hyperspectral imaging in the 460-1020 nm range as a non-destructive tool to achieve these aims in Persimmon cv. 'Rojo Brillante' which is an astringent cultivar. Fruit were harvested at three different stages of commercial maturity and exposed to different treatments of CO 2 (95% CO 2-20 ºCfrom 0 to 24 h) in order to obtain fruit with different levels of astringency. Partial Least Square (PLS) based methods were used to classify persimmon fruits by maturity and to predict flesh firmness from the average spectrum of each fruit. The results showed a 97.9% rate of correct maturity classification and an R 2 P of 0.80 for firmness prediction with only five selected wavelengths. For astringency assessment, as our results showed that the soluble tannins that remain after CO 2 treatments are distributed irregularly inside the flesh, a model based on PLS was built using the spectrum of every pixel in the fruit. The model obtained an R 2 P of 0.91 which allowed the creation of the predicted distribution maps of the tannins in the flesh of the fruit, thereby pointing to hyperspectral systems as a promising technology to assess the effectiveness of the deastringency treatments that are usually applied before commercialising persimmons from astringent cultivars.

Multispectral Imaging for Plant Food Quality Analysis and Visualization

“…Regarding persimmon firmness, 3 (580, 680, and 1050 nm) and 5 (570, 590, 680, 710, and 990 nm) feature wavelengths were, respectively, chosen by PCA and SPA. These 2 groups of wavelengths were then used as independent variables in PLSR models to predict persimmon firmness, yielding the same accuracy ( R 2 P = 0.80) (Munera and others , ). NIR multispectral imaging (900 to 1700 nm) has also been extensively used to measure the physical attributes of plant foods.…”

“…As shown in Figure , it is observed that the soluble dietary fiber content in celery decreases with an extension of storage time, while the insoluble dietary fiber content keeps increasing (Yan and others ). For visualizations of caffeine content in coffee bean, tannins in persimmon, and starch content (SC) in potato and sweet potato tubers, PLSR models in combination with wavelength selection methods (such as RF, SPA, and FMCIA) performed the best (Zhang and others ; Munera and others ; Su and Sun ). In a recent study by Su and Sun (), the simplest FMCIA‐MLR and FMCIA‐PLSR models that were developed with only 6 feature variables resulted in very high R 2 P of 0.962 and 0.963 for detecting tuber dry matter content (DMC) and SC, respectively.…”

“…Linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), and SVM models with 3 characteristic wavelengths (580, 680, and 1050 nm) that were selected by PCA were proposed to grade the persimmon fruits into 3 ripeness stages, with correct classification rates of more than 92% for these 3 classifiers (Munera and others ). Also, 5 optimal wavelengths that were chosen by SPA allowed a 97.9% rate of correct classification of persimmon maturity (Munera and others ). Besides, the ability of multispectral imaging to classify tea leaves by storage periods (from the year 2004 to 2013) was tested, and the higher classification accuracy of 97.5% was determined by the BPNN model (Xiong and others ).…”

Prediction of the level of astringency in persimmon using visible and near-infrared spectroscopy

“…Techniques based on the spectrum analysis, like hyperspectral imaging, have been widely used for the qualitative and quantitative determination of different properties in fruit (Lorente et al, 2012). Munera et al (2017bMunera et al ( & 2017a analysed the astringency and the internal quality of persimmon using hyperspectral imaging, which has the advantage of obtaining both spectral and spatial information. However, one of the most common techniques currently used in food chemistry is near-infrared (NIR) spectroscopy, as it is non-destructive, inexpensive, rapid and reliable (Nicolaï et al, 2007;Vitale et al, 2013;López et al, 2013).…”

Models to improve the non‐destructive analysis of persimmon fruit properties by VIS/NIR spectrometry

“…More specifically, this technique has been used recently to evaluate astringency, tannin content and internal quality of intact persimmons …”