Signal Optimization Approaches on the Prediction of Apples Firmness by Near Infrared Spectroscopy
To evaluate the firmness of Fuji apples that grown in Shanxi province of China, the partial least square regression (PLSR) model was established from the diffuse reflected Fourier Transformed Near Infrared (FT-NIR) spectroscopy. The NIR spectra signal optimization was performing with dif-ferent spectral pre-processing methods, including Direct Orthogonal Signal Correction (DOSC), Standard Normal Variate (SNV), derivative method and their combinations. The homologous models performances were compared to select the optimum signal extracting method. The results showed that the relative standard deviation of prediction (RSDp) was reduced from 16.65％ to 14.82％ after SNV processing. Also it can be improved through the first and second derivative methods when window threshold value was selected at 29 and 51, respectively. Especially, the RSDp of the first derivative model can reach the accuracy at 14.57％. DOSC pre-processing was effective to improve the model performance, in which the DOSC index was selected as 1 and the tolerance factor was set as 0.02. The DOSC can not only decrease the optimum primary number from 5 to 1 and simplify the model, but also increase the model correlation coefficients from 0.753 to 0.817. Moreover, the RSDp value of DOSC model was decreased to 14.08％. Moreover, the combinations of several pre-processing methods were applied to further improve the model quality. Especially, the 1st derivative combine DOSC treatment has the most excellent performance with the RSDp of 13.68％. It was concluded that the 1st derivative, DOSC, or their combinations could effectively filter off the back-ground and extract useful information of apples firmness, and finally improve the model predictive ability. NIR spectra combined with signal optimization approaches could achieve the requirement of NIR non-destruction determination of apples firmness.
Near Infrared Spectroscopy Signal Optimization Apple Firmness Direct Orthogonal Signal Correction
Bolin Shi Lei Zhao Houyin Wang Dazhou Zhu
Food and Agriculture Standardization Institute, China National Institute of Standardization, Beijing National Research Center of Intelligent Equipment for Agriculture, Beijing 100097, P. R. China
国际会议
南昌
英文
1062-1068
2010-10-22（万方平台首次上网日期，不代表论文的发表时间）