Research On Grain Quality Inspection System Based On Machine Vision

The quality inspection of g rain is very important in agricultural production.At present,the detection of grain quality is generally manual visual method.The reliability of this method is not strong,it cannot measure an accurate pass rate,the reliability of grain quality detection is not high,it is difficult to meet the evaluation requirements of mildew or defect in the process of grain acquisition.Although it is possible to use advanced equipment such as photoelectric color selectors,they are expensive and difficult to popularize.Corn,soybean occupies the pivotal position in our country grain production.Therefore,this paper designs a quality detection system for grain(taking corn and soybean as examples)based on image processing and pattern recognition.This system mainly collects images through industrial cameras,realizes image processing and pattern recognition through the machine vision software HALCON programming,and can quickly and accurately detect the grain quality level.The main work of this paper is divided into the following parts:1.In order to obtain high-quality images of corn and soybean,select the appropriate shooting camera model in the hardware part of the system,adjust the lens focal length and optimize the lighting scheme.2.In the image pre-processing stage,gaussian filter and median filter are used to denoise the image and enhance the contrast,and the image is enhanced by the anti-sharpening mask method.3.In terms of image segmentation,the static threshold segmentation method is mainly used to extract the edges.4.Multi-layer perceptron(MLP),gaussian mixture model(GMM)and support vector machine(SVM)were used for feature classification of corn and soybean.The accuracy of each classifier was above 90%,and the best classifier was selected.5.Grade the corn and soybean according to the national quality grading standard.