Research On Aquatic Freshness Near Infrared Spectroscopy Rapid Detection Method

Rapid detection of aquatic product freshness based on near infrared spectrum technology can make significant and valuable contributions to ensure quality and safety of aquatic products. In rapid detection of aquatic product freshness based on near infrared spectrum technology, the key is dimensionality reduction and modeling of spectral data. In this paper, the specific test object is Penaeus vannamei Boone, this paper studies spectroscopy dimensionality reduction and modeling methodologies in quick and nondestructive spectroscopy of fresh aquatic products based on near infrared spectral. First of all, this paper collects original near infrared spectral data of different freshness Penaeus vannamei Boone by near infrared spectrometer.Then, by using method of principal component analysis and linear discriminant analysis, this paper reduces the dimensionality of the spectral data, and studies different dimension reduction methods for predictive modeling effect freshness differences to get the optimal dimension reductive method. Finally, combining with sparse representation classification algorithm and support vector machine method, this paper establishes the quantitative relationships model between spectrum data after descending dimension and freshness indicator —volatile basic nitrogen(TVB-N), it also study the different modeling method difference on the accuracy of freshness prediction to get the optimal predictive modeling method. Experimental results show that, this paper can accurately forecast Penaeus vannamei Boone freshness index by using principal component analysis to reduce the dimension of the spectrum and using support vector machine to establish the freshness index prediction model. In summary, spectral dimensionality reduction and modeling method provide an effective way for the rapid detection of aquatic freshness.