Study On Vehicle-mounted Classification Equipment For Post-harvest Navel Orange

Navel orange,which is a typical dual-purpose fruit for both medicine and food,is rich in nutrition and sweet in taste.China is a country that grows a lot of navel oranges,and its planting area and output rank the highest in the world.More than 90%of navel oranges in our country are used for fresh food,and this consumption structure promotes the development of navel orange classification in the direction of automation.At present,individual farmers are the main growers of navel orange in our country,and the classification method is still based on manual classification.According to the requirements of modern agriculture,fruit will gradually be graded in the field for quick classification and processing.Therefore,there is an urgent need for vehicle mounted or mobile convenient grading equipment to achieve field grading,which requires the classifier to be compact,easy to disassemble and easy to maintain.This research will develop a kind of navel orange classification equipment based on machine vision detection.By the design of hardware system,the miniaturization of the classification equipment will be realized,which lays the foundation for the vehicle mobile operation.The main research contents of this paper are as follows:1.Design and research of transport,turnover and unloading device for navel orange.According to the appearance characteristics of navel orange and the motion requirements on the roller,the bus equation(f(7)x(8)(28)0.937x~2(10)33~2)of the conveyor roller is derived.Different sizes of navel orange can be rolled on the roller with similar angular speed.According to the needs of miniaturization of the unloading device,a discharging device consisting of stepping motor and lever is designed.The thickness of the spring piece is determined to be 1mm and the length is 82mm through the experiment.The width of the spring piece is determined to be 30mm by calculation and the corresponding unloading operation time is 150ms.Through storage experiments,it was verified that the device would not cause mechanical damage to the navel orange after two times classifications.2.Design and research of navel orange image acquisition device.According to the requirements of image acquisition,the model of industrial camera,lens and external trigger is determined.According to the effect of light test,the light source is white LED strip light source,and the lighting mode is diffuse light.According to the requirement of miniaturization of equipment,a light source box for image acquisition is designed,which is 500mm long,600mm wide and 500mm high.3.Study on the classification method of external quality of navel orange.The quality of navel orange was detected by machine vision on line.Comparing the method of minimum circumscribed rectangle,projection area method and the ellipse fitting method,the results show that the ellipse fitting method is the most accurate in detecting the equator diameter of the navel orange at the second station.The average absolute error,standard deviation and average relative error of the three stations are all the smallest,which are 1.80mm,1.35 mm and 2.19%respectively.The characteristics of navel orange image in the color space of RGB and HSV were analyzed.The hue value of H component was used as indicator of navel orange color classification.The threshold value of first class fruit and common class fruit was 19.7,the threshold of common class fruit and outer class fruit was 26.8,and the accuracy of classification was 92.2%.The navel orange grayscale characteristics of all kinds of defects are analyzed.The local threshold method is used to detect the navel orange defect.The improvement measures are put forward in view of the defect detection problem in this method.The iterative method is used to detect the navel orange defect,and the optimum number of iterations is determined to be 2 times.4.Study on the calculation method of navel orange defect area.According to the fruit shape characteristics of navel orange,the method of calculating the defect area based on the sphere belt method is improved,and a new method of calculating the area of defects is put forward.The surface defects of the navel orange can be regarded as a set of long and narrow defects of one pixel width and parallel to the equatorial diameter.At the same time,each single pixel width strip defects are considered as part of the side of a circular truncated cone with a high one pixel,the area of the defect is calculated by calculating the ratio of the defect to the side of the circular truncated cone.Compared with the test results of the ball band method,the accuracy of the circular truncated cone method is improved by 9.34%,and the maximum average error of the improved method is 4.38%,which satisfies the requirement of the defect detection precision.5.Design of the software system for the navel orange grading equipment.PC terminal program and operation interface were written by VS2013+MFC.Image processing called the library function provided by HALCON10.0.The motion control of conveyor and unloading device is realized by PLC,and the signal transmission between PC terminal and PLC is realized through serial port communication.The software system is tested online.The rate of single line transmission line is up to 4 per second,and the accuracy rate of navel orange classification is 91.43%,which can meet the requirements of application.