Research On The Method Of Apple Automatic Grading Based On Computer Vision

Computer vision technology as a way to achieve automation and modernization industrial and agricultural production, has caused the concern of many researchers when that was born, it is the hot research topic in the areas of artificial intelligence, and the rang of it’s application is very wide, such the industrial control, the biomedical, the agricultural engineering, the military and some other fields. This paper study the methods that detects the quality of fruit and grading it from the results that have detected at the real-time, and backed up the knowledge of the standards of the fruit grading, to achieve automation classification of green apples as example, with the application of computer vision technology. This research results has some significance of social and economic for improving our fruit competitiveness in market.The focuses of this study are solving these two key issues in the automatic grading of apple, they are the methods of detection the apple’s size and the detection the apple’s surface defect, and proposed the method or theory that can extract the outline edge of apple’s which is completed and the method of surface defect inspection, complete an system that can realize the grading of apple that automated, that all based on the computer vision technology.The main contents of this paper and the results have achieved are as follows:1.We study some methods of image preprocessing, such as the method of noise remove, image segmentation, edge extraction, and so on.2.This paper proposed the method that combine the method of improved canny and the method of equivalent diameter to measure the size of apple on the aspect of apple size detection. Through comparison some methods, such as the method of minimum enclosing rectangle, the largest fruit axis method, the method of fruit size and some other methods of size detection we found that the computation of the minimum enclosing rectangle is too large, and the placement of the largest fruit axis method is too strict, and the method of fruit size not only have a large computation, but also have a high requirement on its placement, and the method of equivalent diameter is the best way to measure the size of the apples, for the calculation of it is small and the precision is high.3.We study the methods that use gabor filtering to extract the features of apple’s surface defect data. Through learning and research those feature data that extracted, achieve the result that classification the surface defects combined with the method of fuzzy clustering.4.This paper design the system of apple automatic grading combined the software of VC++6.0, include some modules such as the image acquisition, image preprocessing, image feature extraction, and so on.It laid the foundation for the agricultural products automatic grading in logistics in future that the research on the apple automatic grading based on computer vision, but it is a system project that very complex, and also involving many fields and disciplines, this work is limited due to time and personnel constraints. The prototype tool is still in the experimental stage of exploration work.The future work still need further expansion and in-depth study.