Document Type : Research Paper
Abstract
Volume estimation of most agricultural products using mathematical methods is not accurate because of their irregular shape. An alternative method is image processing, which can be used in continuous sorting machines. The objective of this study is to find a practical method for volume estimation of potatoes using image processing. Fifty potatoes (Marfona variety) were selected as experimental samples. An image was captured from two perpendicular views of each potato sample using a digital camera and a flat mirror. The image was then processed with MATLAB software and the potato's dimensions were measured based on the edge position in the image index matrix. After imaging, the potato volume was estimated by two methods. In the first, volume was determined using an experimental equation based on three perpendicular diameters. In the second, the sum of calculated section volumes from image division into smaller truncated cone sections with elliptical bases was used. The size of the three perpendicular diameters and the length of the truncated cone sections were measured using image processing. After measuring the actual volume of the potato samples by water displacement, the error of the estimated volume of both methods was calculated and compared. Results showed that the image division method using 64 sections estimates potato volume more accurately (8.15% error) than the experimental equation method (20.5% error). Thus, the image division method can be recommended as a practical method for potato grading based on volume.
Keywords
)