عادل بخشی پور

Plum is one of the nutritious and popular fruits in Iran. Depending on the quality of the fresh fruit before harvesting and the drying process, different quality grades of this fruit are produced as dried plums. In this research, a computer vision system and machine learning algorithms were used to classify dried plums into three different quality grades. Different color, shape, and texture features were extracted from the images of dried plum samples, and were used separately and in combination with each other for developing classification algorithms of Multilayer Perceptron (MLP) neural networks, Support Vector Machine (SVM), Linear Discrimination Analysis (LDA) and Decision Tree (DT). In order to reduce the number of features and to extract more important features, the Correlation-based Feature Selection (CFS) method was used. Results showed that the combination of different image-extracted features increases the classification accuracy, compared to individual color, shape, or texture features. In this regard, the Random Forest (RF) DT model using the combination of image features and CFS feature selection algorithm had the highest classification accuracy in the training and evaluation stages. The values of Root Mean Squared Error (RMSE) and accuracy of this model were obtained equal to 0.1958 and 93.75% in the training stage, and equal to 0.2110 and 91.67% in the evaluation phase, respectively. Considering these performance parameters and the nature of machine vision systems, the results of this research can be used to develop an accurate, fast, and inexpensive system for the quality grading of dried plums.

The current research used a hybrid conveyor belt drying system, including a solar water heater and solar-powered infrared lamps, to dry rosemary leaves. Its performance characteristics were evaluated regarding drying time, total energy consumption, solar energy ratio and energy efficiency. In each experiment, 300 g of freshly harvested rosemary leaves at a drying air temperature of 50 °C and different values of inlet air velocity (7, 8 and 9 m/s) and infrared lamp power (0, 150 and 300 W) were fed into in the drying system. The optimum point for the working of the drying system was determined using Response Surface Method (RSM) by selecting the appropriate values for the inlet air velocity and infrared power based on these criteria: to minimize drying time, to minimize total energy consumption, to maximize energy efficiency, and to maximize solar energy ratio. Accordingly, the selected optimum point included infrared power of 300 W and an air velocity of 8.85 m/s. Under the defined conditions, the performance characteristics of the drying system, including drying time, total energy consumption, energy efficiency and solar ratio, were equal to 149.13 minutes, 3868.91 kJ, 52.37% and 0.75, respectively. According to the findings of this study, it can be concluded that using a developed drying system can significantly reduce the need for fossil fuel resources and lead to an acceptable performance efficiency in the drying process of medical plants.

Harvest at inappropriate time is the most important factor in reducing the fruit quality, marketability, and postharvest life of kiwifruit. In this study, the effect of different harvest times on fruit quality, nutritional value, and organoleptic characteristics of two novel kiwifruits of yellow and red flesh was investigated. Fruits from promising genotypes were harvested at 130, 140, 150, 160, 170, and 180 days after full bloom (DAFB). The results showed that fruits harvested after 180 DAFB showed higher weight, length, diameter, dry matter, soluble solids content, and total soluble sugars. Meanwhile, titratable acidity was lower than fruits harvested earlier. The values of soluble solid content ranged from 6.80 to 11.66° brix in golden kiwifruit and from 6.90 to 13.70° brix in red flesh kiwifruit. With delayed harvest, total phenols, ascorbic acid content, DPPH and FRAP antioxidant capacity and hue angle in outer and inner pericarp decreased in both genotypes. Ascorbic acid content in the red-fleshed genotype decreased from 90.34 to 75.57 mg/100 g and in the golden genotype from 74.83 to 50.62 mg/100 g. Fruits from the red-fleshed genotype had the higher antioxidant capacity and nutritional value compared to the golden genotype. There was a positive significant correlation between ascorbic acid content and antioxidant capacity measured by DPPH (r=0.98) and FRAP (r=0.95) methods. In conclusion, novel golden and red flesh kiwifruit genotypes reached the minimum soluble solid content and hue angle for commercial harvest with favorable organoleptic properties only at 170 DAFB and the delay in harvest increased the organoleptic properties in both genotypes. However, the content of their antioxidant capacity was slightly lower compared to the early harvest.

Powdery mildew is one of the most important diseases of cucurbits in the world and Iran. The disease is found in cucumbers, melons, cantaloupes, squash, and watermelons in both farm and greenhouse conditions and causes great damage. Cultivating resistant cultivars or chemical control with fungicides can be used to control this disease. In this study, 28 different genotypes of hull-less seed pumpkin were studied and compared for powdery mildew resistance in potted conditions in a greenhouse. To achieve this, the genotypes in the greenhouse were sprayed with powdery mildew spores, and when signs of powdery mildew appeared through infection on the plants, the severity of the disease was determined by image processing using a program written in MATLAB software. The mean comparison showed that the lowest level of infection was related to lines 11 and 23 and hybrids of 7×16, 2×16, and 2×23. The highest negative general combining ability was related to line 11 and the highest negative specific combining ability was related to hybrids 7×14, 2×16, 2×23, and 11×14. The highest negative heterosis compared to the high parent was related to 11×14, 2×23, and 2×16 hybrids. Due to the significant values obtained for specific combining ability and significant heterosis in hybrids, the role of dominance variance in controlling this trait is prominent, and hybrid production is recommended to improve resistance to powdery mildew.

This study was aimed to investigate the best features extracted from images to determine the best technique for qualitative classification of green tea by using meta-heuristic algorithms. Five different classes of green tea were evaluated according to the standards of the Institute of Standards and Industrial Research of Iran. After receiving the images of different green tea classes in the computer, 6 square blocks were extracted from each of the original color images. These image blocks were transformed from RGB to gray scale images. One-level discrete Haar wavelet filter was applied to the gray images and 4 wavelet subimages were obtained. Co-occurrence matrices were calculated for each wavelet subimages and 17 common texture features in the image textural studies, were extracted from subimages (totally 68 texture features for each block image). By using principal component analysis, 8 feature components were produced from the original features and used for the separation of 5 groups of green tea. The results showed that algorithms of artificial neural networks, support vector machine and decision tree were capable of qualitative classification of green tea with high accuracy. However, Bayesian network did not have acceptable performance. According to the evaluation statistics, the multilayer perceptron artificial neural networks (with Kappa statistic, root mean square error and classification accuracy of 0.90, 0.42, and 99.17%, respectively) was the best classifier. Based on the results of this study, the use of machine vision and texture features extracted from image wavelet subimages is a suitable technique for the qualitative classification of green tea.

Manual harvesting of saffron as a laborious and exhausting job; it not only raises production costs, but also reduces the quality due to contaminations. Saffron quality could be enhanced if automated harvesting is substituted. As the main step towards designing a saffron harvester robot, an appropriate algorithm was developed in this study based on image processing techniques to recognize and locate saffron flowers in the field. Color features of the images in HSI and YCrCb color spaces were used to detect the flowers. High pass filters were used to eliminate noise from the segmented images. Partial occlusion of flowers was modified using erosion and dilation operations. Separated flowers were then labeled. The proposed flower harvester was to pick flowers using a vacuum snapper. Therefore, the center of the flower area was calculated by the algorithm as the location of the plant to be detected by the harvesting machine. Correct flower detection of the algorithm was measured using natural images comprising saffron, green leaves, weeds and background soil. The recognition algorithm’s accuracy to locate saffron flowers was 96.4% and 98.7% when HSI and YCrCb color spaces were used. Final decision making subroutines utilize artificial neural networks (ANNs) to increase the recognition accuracy. A correct detection rate of 100% was achieved when the ANN approach was employed.