Research On Color Recognition Of Sugar-Smoked Chicken Thighs Based On Machine Vision Technology

Machine vision technology is a technology that uses an image sensor to obtain an image of a sample,then converts the image into digital information,and uses a computer to simulate human discrimination criteria to identify the image,thereby making an objective assessment.It has a fast detection speed and no damage,no pretreatment,high accuracy,and many other advantages,is a very suitable method for the application of smoked chicken thigh color research.This article takes the smoked chicken thigh as a research object.Based on the machine vision technology,the smoked chicken thigh color recognition model is screened and the algorithm is used to construct and filter the identification colourimetric card.Then,the regression is established by the correlation between the color parameters and the physical and chemical indicators during storage.The model quickly detects the shelf life of smoked chicken thighs.The specific research content is as follows:1.Explore a machine vision model that can quickly and accurately identify all the colors produced by sugar-smoked chicken thighs during the smoking process.Based on machine vision technology,an Xception-CNN model was constructed to identify the color of smoked chicken thighs.At the same time,ResNet-50,Inception and traditional CNN were used to compare and analyze the effect of Xception-CNN model on the color of smoked chicken thighs.After collecting and preprocessing the images,a total of 4352 smoked chicken thigh images of different colors were obtained as test samples of the 4 models.3482 of them were randomly selected as the training group,and the remaining 870 as the test group.The results show that the average recognition accuracy of the four models are: Xception-CNNis92%,ResNet-50is91%,Inceptionis89%,traditional CNNis87%,and test time: Xception-CNNis1.36 s,ResNet-50is0.81 s,Inceptionis0.98 s,traditional CNNis2.48 s.The Xception-CNN model has the highest color recognition accuracy for sugar-smoked chicken thigh images,and has a high recognition accuracy rate for possible smoked chicken thigh colors.The average recognition accuracy rate is92%,and the test time is slightly longer than that of the ResNet-50 model and The Inception model,but lower than the traditional CNN model,takes only 1.36 s to complete the recognition.2.Using machine vision technology combined with three algorithms,three types of sugar-smoked chicken thigh color comparison cards were established,and the accuracy of the three types of color comparison cards was preliminarily verified,and the color matching cards most suitable for color recognition of smoked chicken thighs were screened.The colourimetric card production results show that all three colourimetric cards show good color gradient changes,which can reflect the color change of smoked chicken thighs with the change of smoking time;the colourimetric card verification results show that the three colourimetric cards The accuracy of the algorithm verification is 87.2% of the mean algorithm,95.1% of the K-Means algorithm,and 96.7% of the K-Means algorithm + image noise reduction processing.The accuracy of the sensory verification is 69.4% of the mean algorithm,and the K-Means algorithmis80.9%,K-Means algorithm + image noise reduction processingis79.2%;the colorimetric card drawing system has good feasibility and the accuracy of colorimetric card recognition is high,of which the colorimetric card produced by K-Means algorithm high accuracy and better sensory attributes can recognize the color of smoked chicken.3.A shelf life identify model based on the color change of smoked chicken thighs under 4℃storage condition was constructed,and the feasibility of the model was preliminarily evaluated.Chicken thigh smoked with white sugar was selected as the research object,and the color index,physical and chemical index and sensory index of smoked chicken thigh under the storage condition of 4℃ were measured.Select the color index of smoked chicken thighs,use image processing algorithms to realize the automatic conversion of RGB,HSI and L*a*b* color spaces,and establish multivariate linear regression through the correlation between the color parameters of smoked chicken thighs and TVB-N and TBA values model.Finally,the image algorithm is used to realize the color-based TVB-N and TBA value image visualization.The results show that the multiple linear regression model has a good fitting effect(R2=0.947-0.993),and can predict the changes of TVB-N and TBA values during storage at the same time.It can be used as a low-cost online non-destructive identify of smoked chicken thigh shelves Period method.