Omid Doosti Irani; Abbas Rohani; Mahmood Reza Golzarian; Mansoureh Shamili; Peyman Azarkish
Abstract
Introduction: The diversity and abundance in quality properties of agricultural products are leading factors to develop non-destructive methods. Machine vision and artificial intelligence are powerful techniques in detection of many physical, mechanical and chemical properties of agricultural products. ...
Read More
Introduction: The diversity and abundance in quality properties of agricultural products are leading factors to develop non-destructive methods. Machine vision and artificial intelligence are powerful techniques in detection of many physical, mechanical and chemical properties of agricultural products. Prior to exporting, fruits are sorted in terms of their shapes, volumes or weights. Non-destructively taste-based sorting can be of importance in terms of markability and application. Artificial Neural Network (ANN) has been introduced as a new method to predict quality parameters such as firmness, total sugar content (TSC) and pH of agricultural products and to grade the products accordingly. Material and Methods: In this research, the quality properties of Mango (Kelke- Sorkh var) were predicted using the combination of image processing and artificial intellect techniques. The mango samples were harvested from the orchard in Minab, Hormozgan province in Iran. The samples were transferred to computer vision lab, room temperature of 24°C and 22% RH. The samples were divided into three groups for temperature treatment. They were kept at three temperature levels of 5°C, 15°C and 24°C (control group) for 48 hours. The sample were then placed in room temperature and were imaged every second day for 14 day period. After imaging, each sample was undergon destructive tests for determining their quality attributes including sugar content, firmness and pH. The images were taken by a digital camera in visible spectrum (Nickon Coolpix p510, Nikon Inc, Japan). The taken images were, then, transferred to Matlab software environment (Mathworks Inc, US) for analysis and processing. The color factors from regions of intrest were extracted from the images in L*a*b* color space. The segmentation of images was performed by thresholding (threshhold value of 0.3) the image of difference between red and blue channels of taken RGB images. The conversion of RGB color space to L*a*b* was done by converting RGB image to XYZ basic color space first and before converting X, Y, and Z basic color components to L*, a*, b* color factors. The L* represent the lightness in the image from black (0) to white (100). In this project, a multilayer perceptron neural network with a hidden layer was used. The optimum number of neurons in the hidden layer was found to be 25. The maximum iterations was set as 1000 and the learning rate was set as 0.001. Results and discussions: The input variables to the network were temperature treatment at three levels (control, 5°C and 15°C), the color factors (L*, a* and b*) and the variations of three color factors across the regions of interest (standard deviations of L*, a* and b*). The output variables were sugar content, pH and texture firmness. The results showed that the accuracy of the ANN model on the prediction of pH, sugar content and firmness were 45%, 85 and 88%, respectively. Although the accuracy of ANN model for predicting pH from color factors was rather low, this model was able to predict firmness and sugar content with highly accurately. The histogram of errors among three ANN models also showed the ANN model for predicting firmness and sugar content performed better than that for predicting pH. The MAPE prediction error were 9.53, 22.74 and 6.14, respectively, for predicting firmness, pH and sugar content. Comparing the results from the network in training and testing stages showed that ANN can be considered as a reliable method for estimating quality factors of firmness and sugar content with high accuracy and estimating pH with rather non-applicable accuracy.
Mahmood Reza Golzarian; Mansoureh Shamili; Omid Doosti Irani; Peyman Azarkish
Abstract
Introduction: Machine vision, which uses image processing techniques, is a branch of artificial intelligence that simulates human vision. These systems can be used for quality control, sorting and grading of agricultural products. Unlike engineering materials, agricultural fruits are living tissues that ...
Read More
Introduction: Machine vision, which uses image processing techniques, is a branch of artificial intelligence that simulates human vision. These systems can be used for quality control, sorting and grading of agricultural products. Unlike engineering materials, agricultural fruits are living tissues that continue living even after they are harvested from trees or bushes. Therefore, the post-harvest process such as handling and packaging need to be done such that they make the least damage to these products (Barchi et al., 2002). Combination of suitable techniques and post-harvest management is required to bring down the waste loss in this supply chain. Fruits are susceptible to receive mechanical damages during harvesting (either manually or mechanized), or in transport or at the time of initial packaging. These damages may cause damage to the internal tissue of the fruit that subsequently causes the internal substances of the damaged cell leave spread out. While eradicating the fruit, the surrounding fruits are also affected negatively.Mangos are sensitive to mechanical and thermal sudden change (Xing & Baerdemaker, 2005). Today, surface defect detection and grading of many fruits including mangos are still performed in many cases with the help of trained workers which is time consuming and cost effective.. Image processing has been successfully used for measurement and calibration of products; it shows also a good potential to be used for assessing the quality of products (Mata et al, 2012). There has been no or very little research on the quality assessment of mangos based on the dark spots on the skin surface of mango fruits.The aim of this study was to detect and identify surface damage in mangoes of Kelk-e Sorkh cultivar using digital image processing as it has higher accuracy and processing speed as opposed to manual detection.Materials and methods:Mango fruits were picked from a garden in Minab in Hormozgan province, in Iran. Sixty samples were selected for imaging. These samples had black spots on the skin surface due to mechanical damages received during harvesting and handling. The imaging was performed in a homogenously controlled lighting condition(in an imaging box)against a blue sheet as background and at 24°C and 22% RH. The images were taken in visible range with a Nikon Coolpix P510 digital camera (Nikon Inc, Japan) of 4928 x 3264 dimensions (16.1 MP resolution). Considering the camera lens’s focal length, the samples were placed 20 cm under the camera’s lens to be in camera’s field of view. The taken images were read and analyzed in Matlab (Ver 2011a, Mathworks Inc, US). The quality of segmentation process, which is an important step in image processing project, affects the quality of information extracted from the objects or regions of interest (ROIs) in the next steps. The images of mangos were segmented from the background using thresholding of the high contrast images of red and blue difference. The optimum threshold value was obtained to be 0.3. Then, the affected and healthy regions of mangos were specified manually in each image. Then, the color features in two L*a*b* and RGB and HSI color models were extracted from each region on the sample surface.Results and discussion: The statistical analysis of these features showed that the accuracies for detecting the surface defects on mangos were 90% and 91.6% using the color factor of G and 0.16*G/0.5R in RGB color space, respectively. However, from the a* data, only 56 samples were correctly classified as damaged. This showed the classification accuracy of 93.33% using this color parameter. The accuracy reached to 100% when the two color parameters of a* and L* was used as an integrated color parameter of 0.16*L-a*. In L*a*b* color space, the influence of ambient light on the color of samples is trivial and much less than that on RGB. This can be the reason for higher classification error when R, G and B color components, which might be due to non-uniform lighting and the existence of highly bright or highly dark points on the surface of samples. According to USDA standard, the ratio of the size of defect region to the size of whole fruit can be used as an indicator for grading mangos (USDA, 2006). In this research, the k-means clustering was used to group the mangos based on their defect region size. The results showed that the mangos could be classified into three categories of grade 1, with the defect size of less than 5% of the total area, grade 2: when the defect region size was between 5 and 15% and grade 3: when the defect area size was more than 15% and less than 25%. By K-means clustering, the samples were grouped in two clusters. The cut-off point between two clusters was found from the ROC curve to be 3.11. The parameter of ROC area was equal to unity, which indicated the high discrimination capability of the clustering model. Conclusion: In this research, after assessing several color factors and their combinations, four color components of a*, green (G), 2G/R and L*-0.16a* were selected and used for classifying mechanically defected mangos and the results were promising. The results of this research and similar ones can provide helpful recommendations in grading mangos considering the higher capability of Hormozgan province in Iran for producing mangos for fresh consumption, being used in high-quality domestic market, being exported to global markets.