Research On Sorting Technology Of Chinese Prickly Ash Based On Shape Features

Chinese prickly ash is a kind of widely used cash crop with high edible and medicinal value.Due to the complexity of material composition and small color difference,it is difficult to distinguish from color in the process of production and processing of color selecting machine.Taking the image of samples collected by the color sorter as the research object,this study developed an automatic extraction algorithm to extract the shape features of the material,extracted the area,perimeter and minimum external rectangle of the material,established a complex feature plotter and formed the feature classification data set.The model parameters of support vector machine and penalty coefficient C are equal to 0.4498 and 1.84 respectively.The maximum depth of decision tree model parameters is 6,and the minimum number of leaves and nodes is 10.Deployment model in C++ environment,the simulation experiments on the sample image and validation features describe the son classification effect,analysis the different effect of classification algorithm model for the classification of samples,the simulation results show that: 1)established based on shape feature contains area,perimeter,circle,rectangle degrees,diameter,length,diagonal,compactness,long axial length,a total of nine characteristics to describe the son of Chinese prickly data set all kinds of don't have a separability of samples;2)classification results of support vector machine model of prickly ash: the net rate of shell is 93.63%,and the loss ratio is 3.1%;The net rate of seed is 98.4% and the loss rate is 8.3%.Classification results of decision tree classification model were as follows: the shell purity rate and loss rate of shell are 93.66% and 2.69%,the seed purity rate of seed is 96.01% and the loss rate is 6.9%.The interface visualization software of the above model is built on QT platform to realize automatic feature extraction and rapid model establishment.It is simple to use and portable to embedded system.The research shows that the sorting technology based on shape characteristics is feasible,the sorting algorithm model developed works normally,and the simulation experiment results are good,which provides technical support for the sorting production of Chinese prickly ash and has a certain application promotion price.