Research On Apple Sorting System And Method Based On Machine Vision

China has a very large apple market.In recent years,apple planting area,apple output and other indicators have shown a rapid growth trend.But the backwardness of apple sorting technology greatly limits the development of apple industry.Therefore,it is of great practical significance to study the method of apple size,disease classification and recognition based on machine vision and the automatic apple sorting system.This paper designs an apple sorting system based on machine vision.The image acquisition,image processing,size grading and diseased apple sorting methods of the sorting system are studied.The preprocessing method of apple image is proposed,and the edge contour of apple is extracted.The detection method of apple diameter and size based on improved Hough transform is proposed,and the size detection of apple is realized.A method of apple disease identification based on improved Res Net101(residual network)and migration learning is proposed.The initial data set of apple disease identification was established,and 6000 images with different diseases were finally obtained through the data enhancement method.Res Net50,Res Net101,IRes Net50,IRes Net101(Improve Residual Network)models were compared and analyzed on the apple data set as the improved Res Net theme network and combined with the migration learning to identify the apple recognition results.Finally,IRes Net101 model was selected as the final main network,The batch size,learning rate and other parameters are optimized.The upper computer image processing software and control system software are developed with the sub function modules of image storage and real-time display,image classification processing,job information display,communication,etc.The working performance of the whole machine is verified through tests,and the accuracy is taken as the evaluation index.The test results show that the IRes Net101 model has a higher recognition accuracy than other models.When the feeding speed reaches 10pieces/second,the sorting accuracy of the device can reach 97%.When the feeding speed continues to increase,the accuracy will decline significantly.This is mainly because the inertia of the hierarchical actuator is sluggish,and the reliability of the device decreases,resulting in the failure to clean up defective apples in time.The main research results of this paper can provide some reference for the further research of apple intelligent sorting technology,and provide some reference for the comprehensive promotion of apple intelligent sorting technology.