Based On Machine Vision Technology Of Walnut Size Grading Online And Defects Picking Out System Design Research

The size of the walnut appearance grading as a connection between walnut harvest and deep processing,that will improve production efficiency and increase the value of the product.Based on the external quality of the walnut in Xinjiang,this paper studied the online grading system of walnut size grading and crack removal based on machine vision.The main research contents and conclusions are:1)A review of the development of agricultural products processing at home and abroad based on machine vision.By reading the literature and research,to understand the status of machine vision development,and its application in agricultural product classification analysis of the case,in view of the existence of some shortcomings,combined with machine vision in walnut grading in the advantages of establishing the technical route.2)According to the characteristics of walnut exterior quality and classification,the image preprocessing method is established.Including the image data reduction,image noise reduction and color channel selection,and the initial realization of the walnut in the image of the edge of the extraction,in order to achieve the goal of segmentation and image understanding to create the conditions.3)Study on walnut size and crack classification based on machine vision.The results show that under level ? =0.01,dynamic calibration method with artificial accurate measurement without significant difference.The low-pixel histogram feature method and the gray-level cooccurrence matrix method are used to describe the characteristics of walnut crack,and the entropy,the consistency,the contrast and the homogeneity of the low pixel level of the walnut are extracted as the eigenvector to train the support vector(SVM)classifier,the classification accuracy of the generated classifier model(XML)is 98.3% and 88.0%.4)Based on the machine vision walnut size grading and crack removal of the hardware system structures.(1)the use of stepper motor to complete the camera height lift,making the machine debugging more convenient;(2)dual sensor image acquisition signal trigger and solenoid valve action signal trigger;(3)dual STC89C52 microcontroller composed of lower computer Control center,complete walnut grade signal access;(4)use of air pump and pneumatic solenoid valve walnut grading action.5)For the size of the classification and crack removal of the host computer and the lower computer software design.Designing of MFC interface interaction PC,to achieve before and after the image display,image acquisition optimization,level display and lower computer signal transmission and other functions.The dynamic on-line test shows that defect removal accuracy rate of 70%,the size of the overall accuracy rate of 93.3%,classification efficiency of 3.47 walnut per second.