نویسندگان
چکیده
کلیدواژهها
عنوان مقاله [English]
نویسندگان [English]
One of the steps in potato harvesting is to separate potato from clod and stone that cannot be entirely done by the harvester machines based on the friction and vibration and must be completed by labors. In the present study, image processing technique was used to detect the potato tubers and separate them from clod and stone. Four hundred potato, 100 stone and 100 clod specimens were randomly selected. After selecting the optimum imaging condition, the image of all specimens was individually acquired. An image processing algorithm was designed for preprocessing and extracting different color and texture features in MATLAB. From the features extracted, nine useful features were selected for classification. To classify the specimens, the support vector machine was used by considering two strategies: 3-way (potato, clod and stone) and 2-way (potato and non-potato). The correct classification rate of 3-way and 2-way strategies was obtained 98.67 and 99 %, respectively with the same mean square error (0.0017). According to this result, a machine vision system can be implemented for separating potato from clod and stone.
کلیدواژهها [English]
)