Automatic Classification and Detection of Oranges Based on Computer Vision

In this study, an automatic orange grading detection method based on computer vision was proposed. First, the images of orange were collected and preprocessed. Secondly, orange and background are segmented by edge detection and image segmentation. Based on the image segmentation, four main features of orange include fruit surface color, fruit size, fruit surface defect and fruit shape were extracted. These features were studied using BP neural network. Finally, the automatic orange grading detection was performed by using neural network. The experimental results showed that the grading accuracy of this grading method was 94.38% and the classification accuracy of the first grade is 100%. Compared with the artificial method, it has the features of high correct identification rate and good real-time performance.