Fruit Quality Inspection And Sorting System Based On Machine Vision

Aiming at the shortcomings of traditional deep learning network model,such as large sample size,slow convergence speed,weak generalization ability and low recognition speed,this paper firstly transforms the original data set into the standard format of size based on Anaconda tool and the resize function in Open CV,Using the image class of tensorflow.keras.preprocessing interface to create an image generator imagedatagenerator to enhance the data set efficiently and form the original data set 1.Based on the bilateral filtering and local threshold processing data set 1,the data set 2 containing the main features of the target contour and internal texture is constructed.Secondly,for these two kinds of datasets,based on the original network model of perception,the improved network model of feature extraction of perception is constructed by using convolution kernel decomposition and construction,batch normalization operation,activation function optimization,double sample dropout layer processing and other methods,which makes the training time of the model shorter,the convergence speed faster,and improves the generalization performance of the model for target recognition Accuracy and rapidity.In view of the problems of traditional PLC programming methods,such as a large number of networks,complex structure,low program stability and slow running speed,this paper first discusses the existing PLC programming methods and programming methods in detail,and makes a careful comparison between each other’s characteristics,and puts forward two indicators to evaluate the programming methods: single network integration degree,logical relationship clarity,Thus,it reveals the logical main line from simple engineering to complex engineering.Secondly,based on the above foundation,through the introduction of a complex engineering case,the improvement and implementation of PLC programming method are elaborated in detail,and verified by simulation,which proves that the improved method has practical production significance and value.Then,based on the output signal of machine vision recognition and the improved PLC programming method,this paper constructs the corresponding manipulator grasping control program by controlling the servo motor mobile manipulator platform and rotating manipulator with S7-200 smart-plc,and configures the corresponding main program window,manual program window and automatic program window with MCGS touch screen.Finally,the experiment is carried out in two parts of machine vision and manipulator control,which proves that the idea and algorithm proposed in this paper have good recognition effect,operation efficiency and stability.