Research of Navel Orange Defect and Color Detection Based on Machine Vision
2014 ◽
Vol 513-517
◽
pp. 3442-3445
◽
Keyword(s):
For the problem of high intensity, low efficiency and poor accuracy in the artificial classification of navel oranges, a detection method is proposed based on machine vision technology. In defect detection, we analyze the color information of navel oranges surface, and obtain surface defect with a proper ratio of R/B and G/B. In the detection of color and luster, we calculate the texture information of the grayscale image, and propose three characteristics such as smoothness R, "consistency" measure U and entropy descriptor e. Finally, a hierarchical model is established based on BP neural network. The test results show that this method can be used for detecting the color and luster of navel orange with a high recognition rate.