درجه‌بندی کیفی انگور پوشش‌دهی‌شده با امولسیون صمغ فارسی و روغن شاهدانه بر اساس صدمات سطحی با استفاده از تصاویر Rgb و دودویی

در پژوهش حاضر روش پردازش تصویر‌ جدیدی بر مبنای دو الگوریتم دودویی و rgb با هدف محاسبه میزان صدمات سطحی به‌طور خودکار و انجام درجه‌بندی کیفی ایجاد گردید. ابتدا انگورها با استفاده از فرمولاسیونی از سطوح مختلف صمغ فارسی (صفر، 1.5 و 3 درصد) و روغن شاهدانه (صفر، 0.075 و 0.15 درصد) و 0.3 درصد گلیسرول پوشش‌دهی گردید و در ادامه با اندازه‌گیری میزان صدمات سطحی انگور در روزهای 1 و 28 به‌صورت دستی (با کمک نرم‌افزار image j)، ضمن بررسی اثر پوشش در قالب طرح فاکتوریل (طرح پایه کاملاً تصادفی)، عملکرد الگوریتم‌های پیشنهادی نیز ارزیابی گردید. در الگوریتم دودویی پس از پیش‌پردازش تصاویر، تصاویر به تصاویر دودویی تبدیل شدند. در الگوریتم rgb، فرآیند با کمک مقایسه آماری بین مولفه‌های رنگی صورت پذیرفت. پس از حذف دم انگور و محاسبه مناطق معیوب با کمک گشتاور تصویر (مرتبه صفر و اول)، در نهایت بر اساس درصد مناطق معیوب به 4 درجه کیفی عالی (کمتر از 5 درصد)، درجه 1 (بین 5 تا 20 درصد)، درجه 2 (بین 20 تا 35 درصد) و درجه 3 (بیشتر از 35 درصد) درجه‌بندی شد. مشخص شد که از کانال‌های r، g و b با مقدار سطح آستانه 0.35، 0.45 و 0.3 می‌توان برای فرآیند تشخیص صدمات سطحی استفاده نمود. نتایج نشان داد که هر دو الگوریتم دودویی و rgb توانستند فرآیند محاسبه میزان صدمات سطحی را با صحت بالایی (به‌ترتیب 97.33 و 98.08 درصد) انجام دهند و براساس نتایج ماتریکس درهم‌ریختگی فرآیند درجه‌بندی نیز با صحت بالاتر از 96.30 درصد انجام گرفت. همچنین مشخص شد که پوشش‌دهی با صمغ فارسی و روغن شاهدانه خصوصاً در سطوح پایین سبب کاهش بروز صدمات سطحی طی دوره نگهداری ‌می‌گردد.

Quality grading of grapes coated with Farsi gum and hemp seed oil emulsion based on surface defects using RGB and binary images

Introduction: Grape is a nonclimacteric fruit with a low rate of physiological activity but is subject to serious physiological and parasitic disorders after harvest and during long term storage (Ciccarese et al., 2013). Currently, Edible coatings have been studied as potential substitutes for conventional plastics in food packaging. Edible coating is a thin layer of edible material formed as a coating on a food product. Edible coating can offer several advantages to the fresh fruit and vegetable industry such as improvement in the retention of color, acids, sugar and flavor components, the maintenance of quality during shipping and storage, the reduction of storage disorders and improved consumer appeal (Antoniou et al., 2015; Cazon et al., 2017; Fakhouri et al., 2015; Galus Kadzińska, 2015). Farsi gum as a novel source of polysaccharides has drawn much attention in a wide range of various fields such as pharmaceutics, food and cosmetics industries. Functional properties of Farsi gum are influenced by its structure and molecular weight (Hadian et al., 2016; Joukar et al., 2017). By inclusion of bioactive compounds in the Farsi gum network the aforementioned impairments could be overcome and moreover, new protective and functional valences could be added. The inclusion of lipidbased component in Farsi gum gives it excellent light and moisture barrier properties. The benefic impact on human health of hemp seed oil is worldwide recognized. A recent study demonstrated the antimicrobial properties of hemp seed oil. Due to their abundance in biologically active compounds, hemp seed oil is promising natural alternatives that may extend the shelflife, microbiological safety and nutritional values of food (Cozmuta et al., 2015; Leizer et al., 2000; Salarnia et al., 2018). Growing awareness of the quality of fruit has necessitated increasing effort to develop rapid and nondestructive methods for evaluating fruit quality (Bhargava Bansal, 2020; Rachmawati et al., 2017; Tao Zhou, 2017; Wu Sun, 2013). The aim of this study was the consideration of image processing application for grape sorting based on visual surface characterize. Materials and Methods: Coating emulsion was prepared using (Farsi gum (0%, 1.5% and 3%), hemp seed oil (0%, 0.075% and 0.15%) and glyceride (0.3%)). grape fruit were coated by immersion in coating dispersion for 5 min. Samples were then allowed to loss the excess coating dispersion. Coatings were developed at room temperature during an hour. Samples were refrigerated at 4± 1°C for 28 days and analyses were performed at days 0 and 28. Defect identification and maturity detection of grape fruits are challenging task for the computer vision to achieve near human levels of recognition. The image acquisition was performed in a homogenously controlled lighting condition. Considering the camera lens’s focal length, the samples were placed 25 cm under the camera’s lens to be under camera’s field of view. The images of grape were segmented from the background using thresholding of the high contrast images via MATLAB software (R2019a, image processing toolbox). The optimum threshold value was obtained to be 0.35, 0.45 and 0.30 for R, G and B channel, respectively. Results and Discussion: The proposed techniques can separate between the defected and the healthy grape fruits, and then detect and classify the actual defected area. Classification is performed in two manners which in the first one, an input grape is classified with two different algorithms (RGB and binary). The Result showed that the accuracies for detecting the surface defects on grape were 97.73% and 96.30% using RGB and binary algorithms, respectively. Proposed method can be used to detect the visible defects of coated grape, and to grade the grape in high speed and precision. Conclusions: The results of this research and similar ones can provide helpful recommendations in grading fruits for fresh consumption. The simplicity and the efficiency of the proposed techniques make them appropriate for designing a lowcost hardware kit that can be used for real applications.