期刊名称：Advance Journal of Food Science and Technology
印刷版ISSN：2042-4868
电子版ISSN：2042-4876
出版年度：2015
卷号：7
期号：8
页码：652-657
DOI：10.19026/ajfst.7.1623
出版社：MAXWELL Science Publication
摘要：The storage period of eggs will affect the freshness, therefore the quality and effectiveness of inspection methods to determine freshness is vital. The transmission spectra of egg samples were obtained by hyperspectral image system. Spectrum data and images of egg samples were extracted by ENVI software. After filtering and de-noising, the sensitive waveband (550~900 nm) of spectrum curve were processed by length function. The spectral values of characteristic wavelengths of 625, 650, 675, 750 and 810 nm, respectively, were selected as the characteristic parameters of spectrum. At the same time, R, G, B components of egg samples' spectral images were chosen as the characteristic parameters of spectrum. After analyzing the 8 characteristic parameters by using Principal Component Analysis (PCA) and Stepwise Multiple Linear Regression (SMLR), four parameters were selected as spectral parameters of egg samples which could represent 97.66% spectral information. Using the four parameters (&lambda650, &lambda675, XR and XG) as the independent variable, Haugh unit prediction model of egg samples was established by Multiple Linear Regression (MLR). The correlation coefficient R of calibration set was 0.925 and the correlation coefficient R of prediction set was 0.908. This suggested that the egg freshness detection with hyperspectra was feasible.
Loading...
