Design And Preliminary Experiment Of Walnut Classifier Based On Area Array Camera

XinJiang is one of biggest walnut producing areas in our country.The walnut produce in XinJiang has rich nutrition and flavor fragrance,and is famous in domestic nuts products because the climate condition is superior.Most of the Walnut after harvest use artificial and mechanical methods to classify,but the classification accuracy of these two types are low efficiency and grading index is single,so it is not conducive to the realization of the aim of the quality in the process of walnut commercialization.In this paper,it is designed for Juglans regia ‘Xinxin2’ walnut size grading system using machine vision technology to separation the walnut size.The main research contents are as follows:(1)Design of image acquisition system.According to Walnut size and image acquisition requirement,image acquisition system is designed and established.The main parts of the system include camera,lenses light sources,light box and other key parts of imaging system using experimental and comparative analysis method.(2)Software design of walnut automatic grading.Using VS2010 programming tools,the secondary development of camera was conducted,including control image acquisition,image display,data transmission,etc.The outline of walnut area of pixels was extracted using image processing program based on Open CV2.4.9,and to get minimum circumscribed rectangle.Then short axis pixels of the rectangle were calculated as the index of classification.Moreover,classification program of lower computer was written using PLC program controller and GXDeveloper8.31 programming tools,and signal transmission between PC and PLC used serial communication which use of the Windows API function to realize.(3)Hardware design and manufacture of walnut automatic classifier.The rotating roller of the conveying state mechanics was designed on the base of the principle of the rolling friction of walnuts and Kinematics and dynamics analysis.Double hollow chain is selected to carry the roller as the main structure of walnut single-row and rotary conveying.Electromotor,reducer,frequency converter were selected as the classifier of power and speed regulating device.Aluminum profile was selected to construct rack of the grading.The prototype sorter of walnut was finally finished by purchasing related equipment,parts processing and installation and commissioning.(4)Preliminary test of the sorter.200 Juglans regia ‘Xinxin2’ were randomly choose to detect and complete artificial classification,and were used to test the grading algorithms and the sorter of walnuton-line.The test results shown that image acquisition,image processing and walnut classification is fairly good with classification accuracy of 90%.