Sorting System Of Seeds And Shells Of The Camellia Based On Machine Vision

Camellia oil is well known for its excellent therapeutic effect.In recent years,cultivated area of camellia had been increased rapidly,which leaded to more output of camellia oil.In order to reduce labor cost,overall automated production was needed more.A method that comprehensively considered multiple image features of camellia,such as color,texture and geometric shape was proposed,which combined with machine vision technology to achieve accurate sorting for shells and seeds.The research contents were as follows.1)According to the granularity of camellia,images acquisition system was designed that based on grooved cylinder,which achieved pristine image samples were collected.The work was as follows.a)The overall structure,grooved cylinder and the control system of images acquisition system were designed.b)According to the needs of the acquisition system,the camera,lens and lighting system required by the machine vision system are reasonably selected.2)According to the requirements of sorting,the target contour needed for the experiment was obtained by preprocessing the original image.The following image algorithms are studied.a)Common image denoising algorithms were studied,and the Gaussian filter was selected according to the test results.b)the image color space was converted from RGB space to HSV space,then the saturation channel image was extracted.c)Extraneous pixels were removed and the target contour was clear and intact by morphology open operator.d)The obtained image was binarized by the maximum between-cluster(OTSU),and finally the maximum contour in the binary image was found.3)The characteristic information of the target contour was extracted,which were used to provide data support for modeling.The works were as follows.a)R,G,B,H,S,V mean values of color feature in the target contour were extracted.Scatter plots of color were drawn and analyzed.b)Roughness and contrast in Tamura were used to characterize the texture difference between the shells and the seeds.Scatter plots of texture were drawn and analyzed.c)Area,perimeter,area of circumscribed circle and circularity of the target contour were calculated,and the scatter plot of those were drawn and analyzed.4)The extracted characteristics were used to build the sorting model.The works were as follows.a)The concept of SVM sorting model was introduced and the formula of its common form was deduced.b)The parameters were optimized by the grid search method,particle swarm optimization algorithm and genetic algorithm,the best optimization scheme of which was chosen to establish model of the separation.5)The structure and process of the sorting system were designed on the basis of the image acquisition system.The works were as follows.a)The structure and process of the sorting system were designed.b)The software of sorting system and the interface of upper computer were designed,and the real-time of online sorting was tested.c)The sorting experiment was designed.The errors were analyzed reasonably which were from the test results and the model test results.In this paper,the sorting of shells and seeds of the camellia was studied in detail,which was used to provide technical,supported for the sorting of hells and seeds of the camellia.