ASABE Technical Library - Abstract

Abstract.The pH value of chicken is closely related to the quality of chicken such as chicken tenderness, color,flavor, water-holding capacity and shelf life etc.. This paper have used visible and near-infrared spectroscopy(Vis-NIRS)to build the mathematical model for predicting the pH value of chicken exactly and non-destructively. With chicken breast meat for the object, the chicken breast meat was refrigerated at 4℃. Using spectrometer got chicken breast meat surface spectral reflectance curves, and pH value was measured by portable pH meter. Several mathematical pre-treatments were applied for the samples. The most reasonable pretreatment methods were chosen according to compare the correlation coefficient (R) and standard error of prediction (SEP). And then using partial least squares regression (PLSR) established a detection model about the pH value between the reflectance spectra and the detection model. Finally, verify the accuracy of the model. The results indicated that the original spectra were pretreated by the per-formance of multiplicative scatter correction, then the selected effective wavelengths were used to develop PLSR model .A total of 9 variables selected were employed to construct the model of partial least squares regression (PLSR) . The correlation coefficient was 0.9342 using the model. The correlation coefficient for prediction pre-treatment with Savitzky-Golay smoothing method to establish regression models to reach 0.9161. The results show that the use of visible and near-infrared imaging characteristics predictive modeling not only is useful for nondestructive determination of chicken pH, but also lay the foundation for the further development of on-line testing system.