Sausage Quality Classification of Hyperspectral Multi-Data Fusion Based on Machine Learning

In this paper, the sausage is taken as the research object. Firstly, the total number of sausage colonies is quantitatively analyzed using spectral data of hyperspectral imaging technology. In order to improve the accuracy of the regression prediction model, the key parameters C and gamma of the regression model are optimized, so that it could accurately and quickly predict the total number of sausage colonies. Then, a method of detecting, classifying and identifying sausage key qualities by combining hyperspectral curve features (spectrum) and hyperspectral image depth features is proposed. CNN(Convolutional Neural Network) is used to extract the features of hyperspectral images. At the same time, Principal Component Analysis (PCA) is used to reduce the dimension of hyperspectral curve data (spectrum). Then, the two feature data are fused, and the machine learning algorithm Support Vector Machine (SVM) is combined with PSO (Particle Swarm Optimization) algorithm to classify sausage quality. At last, the experimental results show that PSO+SVM could be used for classification with 99% accuracy after the fusion of spectral curve features and image features. Moreover, the experimental results give the feasibility and effectiveness of this method.