مقایسه خواص فیزیکوشیمیایی انار سالم با انار سرمازده

Abstract
Non-destructive estimation of various physicochemical properties of food such as fruits and vegetables will create a tremendous change in the food industry because it is non-destructive, online, and most importantly, fast. A number of internal characteristics that are of interest to consumers are the content of soluble solids content (SSC), acidity of titration (TA), PH and texture. Therefore, the aim of this research is to compare the physicochemical properties using spectral data, so that if it is significant, the next step can be taken for the non-destructive estimation of the properties. First, 70 healthy and frozen pomegranates were prepared and labeled, and spectral data were extracted using a reflection spectrometer in the range of 900-1700 nm. Then each sample was subjected to destructive tests to measure pH, acidity (TA), soluble solids content (SSC) and firmness. The results showed that, firstly, significant differences were observed between the healthy and frozen classes among all the investigated variables, including spectral data, pH, TA, SC and Firmness. Then clustering was done and the number of healthy and frozen classes that were correctly clustered was extracted. 3 cases could not be placed in any cluster according to their specific class. 66 cases from the healthy class and 62 cases from the frozen class were placed in the correct cluster. The total correct classification rate was 91%.


Introduction
Fruits are among the foods that are rich in vitamins and their consumption is popular among people all over the world. According to the water, air and soil of each region of the globe, different fruits grow. Therefore, in order to distribute these fruits inside the countries or for export to other countries, standards must be taken into account, otherwise, damages will be caused to them in the post-harvest stages and it will cause their quality to decline.
Many researches have been done to measure the external quality of fruits, the results of which are the production of different sorting and grading devices (Blasco et al., 2003; Leemans et al., 2002; Kondo et al., 2000). These methods were all non-destructive methods, that there is no need to destroy the product to determine their external quality. Unlike external quality measurement methods, most of the internal quality measurement methods of fruits are destructive, time-consuming and expensive (Liu et al., 2010). In recent years, various researches have been conducted to predict the chemical characteristics of fruits that determine their internal quality (Arendse et al., 2017). Various methods have been proposed for non-destructive quality inspection, among which are near-infrared spectroscopy (Nicolai et al., 2007), multispectral and hyperspectral imaging systems (Gowen et al., 2007), nuclear magnetic resonance imaging (Marcone et al. al., 2013; Zhang and McCarthy, 2013), X-rays (Donis-González et al., 2014; Magwaza and Opara, 2014). In a research conducted by (Oliveira-Folador et al., 2018), using two near-infrared and mid-infrared spectroscopic methods, they proposed a quick method to evaluate the quality of passion fruit. 130 samples of passion fruit were used for the experiment. Finally, using linear partial least squares regression analysis, they presented models for predicting the chemical properties of fructose, titration acidity, vitamin C, content of soluble solids, sucrose and glucose. The coefficient of determination of these models was in the range of 0.74 to 0.95.Bizzani et al., 2017 presented a non-destructive method to estimate the strength, skin thickness and total pectin content. They used partial least squares regression models of time-domain magnetic resonance spectroscopy, near-infrared and mid-infrared spectroscopy data to predict these characteristics in fresh Valencia oranges. The results showed that NIR and MIR had the best PLSR models for predicting orange firmness with Pearson correlation coefficients of 0.92 and 0.84, respectively.
The purpose of this article is to compare the physicochemical properties of healthy and frozen pomegranates. Because if the results are successful, it will help online systems in non-destructive estimation of the physicochemical properties of agricultural products, which is both fast and accurate.


Methodology
First, 70 healthy and frozen pomegranates were prepared and labeled, and spectral data were extracted with the help of a reflection spectrometer (Iman Tajhiz Co., Kashan). The spectrometer works in the spectral range of 1700-900 nm. Then each sample was subjected to destructive tests to measure pH, TA, SSC and firmness.
The textural characteristics of pomegranate fruit include firmness, cohesion, and elasticity. Among these characteristics, firmness is the most known and important tissue characteristic, which is measured with a pressure measuring device in Newton units.This test will be done by means of a steel rod with a standard diameter of 8 mm, which is connected to a hardness tester. The force corresponding to the maximum penetration value (penetration force) will be considered as a firmness index. Penetration test in the center of the fruit will be done at several points with equal distance on the periphery of the fruit and after removing a piece of the fruit skin.
SSC is measured by an optical refractometer in terms of percentage by pouring one or two drops of fruit juice on the glass plate of the refractometer. To measure the pH of fruit juice, a digital pH meter will be used at ambient temperature. A quantity of juice is poured into the beaker and the pH value is measured by inserting the electrode into the juice.
Conclusion
According to the results of Table 1, significant differences were observed between the healthy and frozen classes among all the investigated variables including spectral data, pH parameters, TA, SSC and firmness. Therefore, non-destructive identification of healthy and frozen classes is possible.
Clustering was done in SPSS software and the number of healthy and frozen classes that were correctly clustered was extracted. 3 cases could not be placed in any cluster according to their specific class. 66 cases from the healthy class and 62 cases from the frozen class were placed in the correct cluster. The total correct classification rate was 91%.In order to make it possible to estimate the physicochemical properties using spectral data, with the help of artificial neural network tools, it is necessary to observe the effective wavelengths in which the differences in properties between the healthy and frozen classes are the largest.