Research on meat species and freshness identification method basedon spectral characteristics
RIS ID
83652
Abstract
This paper proposed a method of meat recognition method based on the artificial neural network ofwavelet denoising. In this study, visible reflected spectra (from 380 nm to 780 nm) of beef and porkwith different freshness were measured with fiber sensor spectrometer. The wavelet multi-resolutionanalysis was employed and the ideal way of decomposing layers was selected to eliminate the burr noiseor abnormal data caused by absorption and scattering spectra signals in optical fiber. Then a kind of witha double-hidden layer was applied to analyze meat spectral reflectance data, and the back propagation-artificial neural network (BP-ANN) was trained by Polak-Ribiere conjugate gradient learning algorithm.The experimental results show that the method can analyse the complex spectrum signals and achievea good identification on the species and freshness of meat.
Publication Details
Wang, F., Jin, L., Zhang, T., Zhang, Y., Ye, J. & Kan, R. (2013). Research on meat species and freshness identification method basedon spectral characteristics. Optik, 124 (23), 5952-5955.