Development of a Real-Time Machine Vision Prototype to Detect External Defects in Some Agricultural Products | ||||
| Journal of Soil Sciences and Agricultural Engineering | ||||
| Article 1, Volume 12, Issue 5, May 2021, Page 317-325 PDF (1.4 MB) | ||||
| Document Type: Original Article | ||||
| DOI: 10.21608/jssae.2021.178987 | ||||
 View on SCiNiTO | ||||
| Authors | ||||
A. R. Mohamed; G. M. El Masry; S. A. Radwan; R. A. ElGamal 
| ||||
| Agricultural Engineering Department, Faculty of Agriculture, Suez Canal University, Ismailia, Egypt. | ||||
| Abstract | ||||
| The automation of agricultural operations not only improves the quality and productivity of agricultural products but also helps in enhancing the national income. Although human sorting and grading are the traditional methods usually used in the postharvest chains, these methods are inconsistent, time-consuming, subjective, expensive and easily influenced by the environment and human fatigue. Therefore, the main aim of this study was to develop a real-time machine vision prototype for sorting and detecting the quality parameters of different agricultural products. The constructed prototype was used for image acquisition and processing. By using the data of color values of all concerned defects, a simple thresholding (min-max method) was developed and employed using Python software. Three types of defects (greening, black spots and scares) in orange, two defects in potato tubers (greening and black spots) and two defects (broken pods and black spots) in peanut were detected using the developed system based on color differences. The system was also used to detect singular peanut pods (half pods containing one internal seed instead of two or three seeds) based on dimensional features. The results obtained in this study revealed that the developed prototype was used successfully to detect the external defects of tested products with reasonable accuracy.The accuracy of defect detection during real-time operations of orange, potato and peanuts were 96.97, 98.50 and 99.09%, respectively. The developed detection method was also very efficient in the classification of the peanut pods into full-size pods and singular pods and with overall classification accuracy of 100%. | ||||
| Keywords | ||||
| Machine vision; real-time; sorting; grading; orange, potato, peanut pods | ||||
|
Statistics Article View: 340 PDF Download: 447 |
||||
