World potato consumption is headed up, growing at an annual rate of 4.5%. Detection of external defects on potatoes is the most important technology in the realization of automatic potato grading stations. A real time system is proposed in this article; HSV color space is used to remove background following image acquisition step. Afterwards, co-occurrence texture features are extracted from the image, and finally three different Neural Networks are trained and validated to select the better classifier for defect detection. Results showed that the Support Vector Machine networks represent a higher performance in the direction of Multi Layer Perceptrons and Radial Basis Function networks for potato classifications.
کلید واژگان :Potato Defect Detection, Real Time (RL), Morphological Operations, Otsu thresholding, Color-space, Support Vector Machines, Multi Layer Perceptrons, Radial Basis Function.
ارزش ریالی : 300000 ریال
با پرداخت الکترونیک