Research On Chicken Wing Quality And Weight Grading Based On Machine Vision

Although China is a big producer of broiler chickens,the research on the classification of broiler chicken products started late,and there is still much room for improvement abroad in terms of technology.Large domestic meat processing enterprises have improved production efficiency through the introduction of high-priced foreign equipment,while most of small and medium-sized enterprises in China still use manual inspection and classification of meat appearance quality on the assembly line.Manual detection is slow and mechanical work is prolonged for a long time.Human eyes are easily fatigued,which can easily cause false detection and missed detection.This research is devoted to the development of fast automatic grading technology of chicken wing quality and weight based on machine vision,applying machine vision technology to the rapid and non-destructive testing of external quality of chicken wings,designing an online weighing and grading system for chicken wings,and improving the efficiency of chicken wing classification.The research focuses on the following four aspects.(1)A method for extracting the congested part of chicken wings.Using digital image processing technology,the collected static chicken wing images are pre-processed by graying processing,threshold segmentation and morphological processing,etc.,to realize the segmentation of chicken wing congestion.(2)Establish a chicken wings quality prediction model.Based on the chicken wings color extraction map,a color cumulative histogram is established for the chicken wings image and the sum of the pixel ratios when the H-S₂ quantization values are 3,4,and 5 is calculated.If the percentage of blood coloration exceeds the set threshold,it is determined to be bloody chicken wings,thus Identification and classification of congested chicken wings and normal chicken wings.(3)Study a method of online automatic weighing and grading of chicken wings.After the weight of the normal chicken wings is detected,the single chip microcomputer receives the detected chicken wings weight information in real time and processes and analyzes them to determine the level of the chicken wings and control the corresponding robotic arm to sweep the chicken wings into the storage box.When the number of chicken wings in the storage box reaches a preset value,the bottom of the box is automatically opened,so that the chicken wings are continuously stored and quantitatively dumped.(4)Establish a set of chicken wing quality and weight grading system platform and conduct test verification.The platform mainly includes three parts:visual processing,automatic weighing and intelligent classification.The sample chicken wings were classified and graded,and the false detection rate,missed rate,classification accuracy rate,classification rate,and classification accuracy rate were tested to verify the operation quality.The prototype test results show that the highest classification accuracy of chicken wings recognition is 100%,the lowest is 98%,and the average accuracy is 98.8%;the highest false detection rate is 2%,the lowest is 0%,and the average false detection rate is1.2%;the average The missed detection rate was 0%.When the conveyor belt speed is0.6m/s,processing speed of chicken wings is 7025 per hour,and the average classification accuracy is 97.8%.The research results are expected to provide relevant theoretical basis and technical support for the development of processing automation equipment for broiler and even poultry segmentation products with independent intellectual property rights in China.