Automated Online Grading System Of Chicken Carcass Quality Based On Machine Vision

Chicken carcass quality grading refers to grading the chicken carcass depend on their quality values,which is an important part in the chicken slaughter production.In the chicken divison line,chicken carcass should be pre-grading before divison part to choose appropriate size chicken to be cut;in whole chicken production line,quality grading even has,a direct impact on the classification of the product packaging and price of the chicken carcasses.At present,the quality grading of chicken carcasses has not been fully automated,and domestic small and medium-sized enterprises still use artificial weighing grading.The workers graded according to the weighing results of scales,whose efficiency is low and the error rate is high.Foreign and domestic large enterprises using automated grading equipment which weighing online in the conveyor belt,after quality data upload computer,the computer will control the grading implementation.Although these equipment own high precision,but repeatedly touching the chicken carcass results in a secondary pollution to the chicken carcass,and causes serious food safety risks.Therefore,it is necessary to study a non-contact quality grading method that will fundamentally avoid the situation of equipment polluting the chicken carcass.In this paper,the non-contact property of machine vision was used to propose a grading method of chicken carcass weight based on machine vision.The images of chicken carcass was collected by industrial camera,then some image feature quantities which closely related to the chicken carcass were extracted from the images by image process technology.The mathematical model of image feature quantities and chicken carcass quality was established,and the quality of chicken carcass was predicted according to the size of features.Contrasting the production of grading standards,the quality of chicken carcass grading was ultimately realized.The main contents and conclusions of this paper are as follows:1.Collected 100 chicken carcass training set of sample images which were implemented graying,filtering,edge detection,binarization,morphology and hole filling process to achieve image preprocessing.Six feature quantities described the quality of chicken carcasses were extracted which were the projection area Sp,carcass length Hp,contour length Cp,chicken length Ap,chicken breast width Bp,chicken breast area Ep.Finally,the image feature quantity is converted from the pixel size to the actual physical size through the image calibration.2.The relationship between image feature quantities and chicken carcass quality was studied.Linear regression analysis and principal component analysis were used to establish the mathematical model between feature quantities and quality.The statistical data of each model were analyzed and compared,and the prediction error of each model is given.According to the quality grade standard,the accuracy rate of the quality grading of each model was tested,and the predicted value and the actual value of the carcass quality were classified and the percentage of consistent judgment was treated as the accuracy rate result.The results show that the average grading accuracy of the grading model is 90%.3.Designed a set of chicken carcass quality grading system,the system is divided into image acquisition,image processing,graded execution of three parts,using the industrial camera to collect images,industrial computer to process image,determine the level,PLC to control pneumatic devices which can remove the chicken carcass.Finally,the quality of chicken carcass online grading was realized.4.The quality grading platform was used to test and evaluate the chicken carcass grading method.Two different experiments were designed to test the classification accuracy of chicken carcass quality grading system at different conveyor chains.The experimental results show that the grading accuracy rate of classification system is 82.3%.