عنوان مقاله [English]
Orange is one of the old and widely used citrus fruits, which, despite the presence of very useful substances and micronutrients, has many properties, especially in blood purification and healing and prevention of many diseases. Oranges are rich in vitamins B, C, iron, calcium, phosphorus, potassium, sodium and copper. Vitamin C, which is one of the important factors in inhibiting cancer, and oranges are the main source of this vitamin. Mechanical shocks are known as effective factor in post-harvest losses. During post-harvest stages, dynamic loads are more effective in causing bruises. Considerable research has been reported on the mechanical properties of fruit tissue, the mechanisms of bruising and the methods of prevention, reduction and detection of bruising. In this research, the aim is to investigate the possibility of using hyperspectral imaging to propose a non-destructive model in identifying bruised oranges. In this regard, first a sufficient number of sound oranges were prepared and their hyperspectral images were taken. Then, bruises were artificially created in them, and hyperspectral images of oranges were taken again 12 and 24 hours after bruising. To perform the destructive test and measure the firmness of oranges, 40 oranges from each class (sound, 12 and 24 hours after bruising) were selected and subjected to the destructive test and the hardness values were recorded. Duncan's test was performed to compare classes. The results showed that in both the destructive and non-destructive tests, there was a statistically significant difference among all the three studied categories, and therefore it is possible to non-destructively identify bruised oranges by hyperspectral images.
Iran's citrus fruits are one of the important domestic products and a source of agricultural income of the country; But unfortunately, with the decrease in competition in the global market. The reason for this is the high cost of production and transportation due to the recent high prices on the one hand and the decrease in the price of export products of competing countries such as Turkey in this period of time.
Improper transportation of fruits causes mechanical damage to them and increases waste. The form of damage depends on the physical and biological structure of the product and the type of load (static, dynamic and fluctuating load). If the transporter is not suitable, during transport, the fruit will shake violently and collide with the surfaces of the transporter or other fruits, and as a result, it will change the shape of its tissues. If this deformation exceeds the biological submission limit, the tissue will be discolored and decayed in a short period of time, and thus the material will be completely destroyed. Rotten products during storage in the warehouse will also endanger the healthy materials that are in contact with them (Sitkei, 1986). The amount of fruit waste in the transportation process has been reported between 30 and 40%. Many studies of Tomato (Olorunda and Tung, 1985), apricot (Holt and Schoorl. 1985), grape and berry (Hinsch et al., 1993), peach (Zeebroeck et al., 2008) have been reported.
Mechanical loads are known as the main and effective factor in post-harvest losses. During the post-harvest stages, dynamic loads are more effective in causing bruises in the products. Fruit bruises often occur during the stages of moving, transportation, and packaging due to impact from the moving parts of machines and other factors. . Because dynamic loads have more effect than static loads in terms of quantity and occurrence (Mohsenin, 2006.). The amount of bruising plays a key role in the separation stage of healthy products and their grading. The loss rate of different fruits is different. For example, apple losses are usually between 10 and 25 percent. But in some varieties, this amount has been reported up to 50% (Hinsch et al, 1993).
Considerable research has been reported on the mechanical properties of fruit tissue, the mechanisms of bruising and the methods of prevention, reduction and detection of bruising. During transportation and storage, excessive static load is applied to some oranges in the fruit container. However, most bruises are caused by dynamic load in the form of vibration or impact.
Recently, optical measurement technology as a potential tool for non-destructive analysis and evaluation of food quality and health has reached a level of development that is available and usable. Hyperspectral imaging was first performed in the late 1970s in the United States of America and has rapidly developed and expanded. The main part of the development of this technology is in 1989, which took place at the same time as the Aviris airborne sensor was built by NASA's GPL center and was able to sample in 224 spectral bands, after which other types of airborne and space hyperspectral sensors were designed and built. .
In various researches, PCA has been used to select several wavelengths that can potentially be used in a multispectral imaging system on the line, which can be used to identify different common defects of oranges using a hyperspectral imaging system in the spectral range of 400 1000 mentioned. The imaging system includes an imaging spectrograph, 150 watt halogen lamps and a CCD camera. The oranges were considered to have insect damage, wind damage, thrips damage, and aphid contamination. In the principal components algorithm used in this research, each principal component is a linear summation of principal images at unique wavelengths that are multiplied by the corresponding (spectral) coefficients.
6 wavelengths of 630, 691, 769, 786, 810, 875 nm in the wavelength range of 550-900 nm and wavelengths of 691 and 769 nm in the visible wavelength range were selected as optimal wavelengths for further analysis (Li et al, 2011).
In this research, first, 40 sound oranges were picked from the citrus garden and placed inside the special fruit foam and carefully arranged in a single row inside the can so as not to get mechanical shocks during the transfer to the laboratory. First, the samples were labeled (Figure 1). Hyperspectral images of all sound samples were taken as a sound categoury, and then artificial mechanical damage was applied to all samples by free fall of a steel ball with a diameter of 6 mm and a weight of 150 grams from a height of 45 cm to cause bruises. Finally, the hyperspectral image of all samples was taken again at 12 and 24 hours after bruising. Therefore, a total of 120 samples were obtained in the form of three categories.
In order to extract spectral characteristics from each sample, a hardware system must be configured. The components of this system are 1- laptop, 2- hyperspectral camera, 3- light source. A penetrometer is a device that is widely used by agricultural and food industry research centers to check and detect the degree firmness of different types of fruit in order to improve quality and determine the appropriate time for storage and transportation. (Figure 3).
Finally, after obtaining the data related to the reflection spectrum of oranges and measuring their firmness, a comparison between healthy and stuffed groups was made by Duncan's test in SPSS software. The purpose of this work was the feasibility of using hyperspectral imaging in the development of a non-destructive method in the detection of bruised oranges.
In order to achieve a non-destructive method in identifying bruises, it was first necessary to analyze at least one characteristic of bruises destructively and evaluate whether the destructive method was able to identify the difference between healthy and bruised categories. For this purpose, the hardness of the oranges of each batch was measured using a penetrometer or a hardness tester (Tables 1-3).
Based on Table 1, we find that the average hardness values for healthy oranges were 0.29 N/cm2. For oranges whose hardness was measured 12 and 24 hours after crushing, respectively, 0.53 and 0.80 N/cm2 were reported. The reason why healthy oranges were more firm than meatballs can be attributed to the crispness of the skin of healthy oranges compared to meatballs. According to Table 2, the difference between healthy oranges and meatballs was significant, based on Table 3 and Duncan's test, it was drawn that this difference prevails between all three categories.
Now that we know that there is a significant difference between healthy and meatball categories in the destructive method, it is necessary to check the significant difference of all three categories in terms of spectral values so that an automatic and non-destructive method can be proposed to identify orange meatballs. As can be deduced from Tables 4 to 6, there is a statistically significant difference at the level of 5% between all three studied categories, and therefore it can be concluded that the hyperspectral imaging method is capable of identifying bruises in oranges.