Research And System Development Of Mutton Freshness Detection Based On Hyperspectral Imaging Technology

Traditional method is applied to detect lamb freshness is not only time-consuming,damage to the sample,there is a certain subjective interference factors,can't accurate judgement of mutton fresh condition,Therefore,it is necessary to seek a rapid,non-destructive and accurate method to promote the development of mutton industry in China.in recent years,with the gradual development and mature of compose a specular like technology,The application of this technology in the field of quality inspection of farm and livestock products has been paid much attention by researchers at home and abroad.Volatile base nitrogen content is important for mutton freshness evaluation index,this article attempts to adopt the model can be nondestructive testing of hyperspectral image technology to predict volatile base nitrogen content in the mutton to achieve rapid non-destructive testing for the determination of the freshness of mutton,the study design based on the technology of compose a specular like mutton freshness determination system,using the Java programming language in the Eclipse integrated development environment for writing software designed specifically for the following research:(1)81 mutton samples from the carcass tenderloin of the sheep were collected from the mutton of xilingol in Inner Mongolia.The content of TVB-N was determined by scalar according to national standard.The original spectral image was corrected by standard normal variate(SNV).The 2D principal component analysis(2D-PCA)was used to reduce the dimension of the original hyperspectral image data,and the feature images at 1257.49nm,1396.19nm and 1736.15nm were optimized.(2)A total of 54 texture and color feature parameters were extracted from the feature images,and the feature parameters with high correlation with the content of tvb-n were screened out.As the input of the artificial neural network(BP-ANN)and partial least squares regression(PLSR)models,the prediction model of the content of mutton tvb-n was constructed.The results show that the bp-ann model has a determination coefficient R2 of 0.86 and a root mean square error of 3.33 for the prediction set,and the PLSR model has a determination coefficient R2 of 0.81 and a root mean square error of 3.96 for the prediction set,so the prediction effect of BP-ANN is obviously better than that of the PLSR model.The results show that it is feasible to quickly and nondestructively detect the content of TVB-N in mutton by using the technique of hyperspectral imaging.(3)Design a software system for predicting mutton freshness.Analyze the system's functional requirements,design relevant functional modules,and realize the basic functions of user login,principal component image extraction,texture feature extraction,feature value preservation,training,and prediction in the software system.The test system has the functions of image dimension-reduction,texture feature extraction,data storage,model training and TVB-N content prediction.The results show that the mutton freshness prediction system can effectively reduce dimension-reduction and accurately predict TVB-N content based on hyperspectral image data,basically achieving the expected function and meeting the design requirements.In this study,a mutton freshness detection system based on hyperspectral imaging technology was designed,which can realize the functions of hyperspectral image data processing and the prediction of the content of TVB-N,providing a new idea for the rapid,accurate and non-destructive detection of mutton freshness,and is of great significance for the development of meat quality detection in the future.