Chilled Pork Freshness Detection And APP Development Based On Shikonin Indicator Label

With the change of meat consumption structure,the market share of chilled meat has gradually increased.However,chilled meat is susceptible to microbial contamination during transportation and storage,which leads to spoilage and deterioration,and even produces substances harmful to human body.Therefore,real-time detection of chilled meat freshness and timely treatment of spoiled meat are of great research value to reduce waste of resources and ensure food safety.This paper focuses on the color information of Shikonin indication label during the storage of chilled pork,determines the chilled pork packaging method applicable to the refrigerated storage of supermarkets,and builds the corresponding image acquisition platform.The chilled pork detection model is established by reading the color information through a series of image processing of the collected label images.And a chilled pork freshness detection software based on Android platform is developed to realize the nondestructive detection and real-time display of chilled fresh pork freshness,providing a scientific detection method for consumers to choose fresh pork.The main research contents and conclusions of this paper are as follows:(1)Build a chilled pork image acquisition device based on Shikonin indicator labels.Sodium alginate,gelatin,glycerin,and Shikonin were used to prepare Shikonin indicator films that show good indication of the freshness changes in chilled fresh pork.The chilled pork was stored in disposable PET plastic boxes with Shikonin indicator labels and white balance calibration cards,and a suitable packaging method for supermarket refrigeration was determined.Through experimental comparison and analysis,a Shikonin indicator label image acquisition device suitable for supermarket environments was designed,and the parameters of the image acquisition device were determined to ensure the quality of the captured images.(2)Establish a freshness detection model for chilled pork based on Shikonin indicator label images.Firstly,gray world method,perfect reflection method and dynamic threshold method were synthetically compared for the label images,and the perfect reflection method was selected as the automatic white balance adjustment algorithm to solve the problem of color deviation.Then,image grayscale,image binarization and region of interest extraction were used to separate the foreground and background of the image,fully keep the label region and remove the useless information.Next,a mask was generated based on the color information of the label area,and the label is cropped according to the position of the mask.The extracted Shikonin indicator label image was then subjected to Gaussian filtering,and its color information was obtained.Finally,22-dimensional color features of the Shikonin indicator labels were extracted,and four models including Partial Least Squares Regression,Support Vector Machine,K-Nearest Neighbor,and Random Forest were established after normalization,respectively.The results showed that among the four color feature models,the Random Forest model had the relatively best prediction effect with 99.65% accuracy in the training set.Moreover,its testing set accuracy,fresh pork recognition accuracy,and spoiled pork recognition accuracy reached 98.94%,98.55%,and 100%,respectively.The fresh meat and spoiled meat can be detected with high accuracy based on the Shikonin indicator label image.(3)Develop the chilled pork freshness detection software based on Android system.Android Studio was used as the development tool and the development environment was set up.The software requirements are analyzed and functional modules are divided,the software operation flow was determined,and the software program and main interface were designed using Android Studio.The software can realize the functions of image input,result prediction and result visualization,and deploy the chilled pork freshness detection model on the client side.