近红外光谱对甜椒果实质地的无损检测

Nondestructive measurement of sweet peppers texture quality based on near-infrared diffuse reflectance spectroscopy

以黄色甜椒为研究对象,建立其近红外漫反射光谱检测果实质地的数学模型.在400～2 500、400 ～1100、400～1 450 nm3个波段内分别建立了甜椒的果肉弹性、回复性和凝聚性定标MPLS模型,并用各波段下最优模型进行预测.结果表明：这3个波段下的定标模型相关系数都很高,但在全光谱下建立的定标模型稳定性最好,所以选取该光谱下的定标模型作为最终的测定模型,果肉弹性、回复性和凝聚性定标集交互验证相关系数（RCV）分别为0.937、0.933、0.932,交互验证标准误差（SECV）分别为0.029、0.013、0.016,预测集的相关系数Rp分别为0.924、0.899、0.922,预测标准误差（SEP）分别为0.026、0.018、0.015,相对残差分别为-0.200、0.068、-0.033.结果说明,甜椒果实质地的近红外无损检测是可行的,果实质地与近红外漫反射光谱具有显著相关性.

This study was to establish mathematical relationship between near infrared diffuse reflections spectros- copy and texture qualities of sweet pepper. Modified partial least squares （MPLS） calibration model of flesh elasticity, resilience and cohesion of sweet peppers were established in three wavelength ranges of 400～2 500 nm,400～1 100 nm and 400～1 450 nm,respectively. Then the optimal model at each band was used in prediction. The results showed that calibration model coefficient was higher at these three bands, but the whole spectrum was selected because it showed the best stability. The flesh elasticity,resilience and cohesion with the correlation coefficient （RCV） was 0. 937,0. 933,0. 932, respectively, standard error of cross validation （SECV） was 0. 029,0. 013,0. 016 N,respectively. Correlation coef- ficient of prediction （RP） was 0. 924,0. 899,0. 922, square error of prediction （SEP） was 0. 026,0.018,0. 015 N, resid- ual error was -0. 200,0. 068, -0. 033 N. The research results indicated that it was feasible for determination of sweet peppers texture qualities based on the significant correlation with the near infrared sneclrnscanv