Jujube Classification Based On Machine Vision And Design Of Sorting Execution Device

Red jujube is rich in nutrients,which has not only edible value,but also high medicinal value.In recent years,the development of jujube industry is flourishing,which brings huge profits to farmers and has become a pillar industry of China's rural economy.However,most postharvest jujube enter the market through the manual sorting,which not only wastes a lot of labor force,but also has low separation efficiency,that leads to the lack of competitiveness of Chinese jujube in the international market.Jujube sorting machine on the market are mostly rolling rod style or roller type classifiers,its grading standard is single and that easily leads to mechanical damage of red jujube.Therefore,it has good prospects to develop a highly efficient and convenient jujube grading detection system.In this paper,the Xinjiang Jun jujube as the research object,we adopt machine vision technology to obtain and process the red jujube images,and then design the red jujube hierarchical execution device.The main contents and conclusions of this paper are as follows:Firstly,the research on jujube image acquisition based on machine vision: The design and construction of red jujube transmission mechanism mainly include the design of roller size,transmission frame and roller shaft,and the selection of the transmission chain.The test results show that the transmission mechanism can transfer most of the red jujube smoothly,and its transmission speed can be adjusted according to the grading requirements.The construction of red jujube image acquisition system hardware includes the installation of black box,light source and industrial cameras.This system can get the images of different surfaces of red jujube,and transfer the acquired images to the computer for subsequent image processing.Through the analysis of the force and motion of the red jujube in the roller conveying mechanism,we get the mathematical model of the red jujube horizontal forward distance from the static to roller line speed by acceleration and the red jujube rotation angle when collecting jujube image by fixed distance interval after the acceleration process.When collecting three red jujube images by fixed distance interval,the 360 degree axial image of red jujube with transverse diameter of 20 ~ 40 mm can be accurately collected.Through compiling single chip microcomputer program to control the red jujube image acquisition,the camera can automatically collect red jujube image by fixed distance.Secondly,study on grading treatment of red jujube image: We extract the jujube image contour by the least squares method,and then classify the jujube according to transverse pixel value of the extracted contour image.We extract the diseased region by scanning H mean of the H component diagram of red jujube image,when the area ratio of disease is0.092,which as grading index,the recognition rate of disease jujube can reach 88.23%.We extract red jujube crack by Otsu's threshold segmentation of I component diagram of red jujube image,then establish crack identification model and the crack identification rate can reach 90.47%.The recognition time of a red jujube image is about 0.2 seconds,which basically meets the requirements of online grading efficiency.In this study,we extract and match the feature points of red jujube image by SURF algorithm,and realize the purification of matching pair by the RANSAC algorithm.We realize the final fusion of the red jujube image by weighted average method,which effectively eliminates the seam problem of image stitching.The jujube image mosaic software is designed by combining Visual Studio 2010 software and Open CV function database,which realizes the mosaic and preservation of red jujube image.However,the stitching time of two red jujube images is about 2 seconds,which does not meet the requirements of online grading efficiency.Moreover,image recognition after splicing is low in efficiency,so we identify the same red jujube samples three times.Thirdly,design and reliability study of jujube hierarchical execution device: Taking into account the classification efficiency,cost,simple control and other factors,we design a push-pull electromagnet execution system.The design of the hierarchical execution device mainly includes the design of hierarchical actuator,hierarchical channel and hierarchical control.The reliability research of hierarchical execution device is mainly to investigate the degree of mechanical damage to the red jujube in the process of classification.In this study,the mechanical properties of red jujube were studied with the installation distance,installation angle and moisture content,the results show that the mechanical damage of the jujube is the smallest when the installation distance of the grading baffle is 20 cm and the installation angle is about 45 degrees,and the mechanical damage to the jujube decreases with the decrease of water content.The control system of the hierarchical execution device realizes the real-time communication between the computer and the single chip through compiling serial communication program.At the same time,shift register 74HC595 for real-time tracking and positioning of jujube ensures the single chip computer can send out the hierarchical control command to pull the electromagnetic relay when the red jujube reaches the grading channel,finally the hierarchical actuator completes the red jujube classification.