Fruit quality grading method based on image processing

The invention discloses a fruit quality grading method based on image processing. The method comprises the following steps: shooting fruit surface images from multiple angles through a camera; pre-processing the shot image, wherein the pre-processing comprises methods of Gaussian filtering, threshold segmentation and the like; extracting color features and size features of the fruit surface, performing region segmentation on defect candidate regions of the fruit surface, and constructing a data set of the defect candidate regions; training a model capable of identifying a defect area and a fruit stem area by using the data set of the defect candidate area; identifying the extracted region part by using the trained network model to obtain feature data of the defect; and classifying fruit quality grades according to a set fruit grading standard by utilizing the obtained characteristics such as the color characteristics, the size characteristics and the defect areas of the fruits, so as to finish fruit quality g