利用近红外漫反射光谱分析技术对种蛋中的无精蛋和受精蛋进行检测。 通过不同波段范围、 不同主成分因子数和不同光谱预处理方法对种蛋类型检测结果的比较分析， 建立了种蛋的定性检测模型。 结果表明： 选取光谱范围为4 119.20～9 881.46 cm-1， 19个主成分因子数， 原始光谱经过SNV+一阶导数+Norris微分过滤的预处理方法， 利用马氏距离建立种蛋定性检测模型， 校正集正确率达到92.50%， 验证集正确率达到91.67%。 该研究为上孵前无精蛋和受精蛋的无损检测提供了新的途径。

The detection of the infertile eggs and fertile eggs by the near infrared diffuse reflectance spectra was proposed. Models based on different band regions range, different principal component numbers and the different spectral pre-processing methods were compared and the optimal calibration model was established. The results show that qualitative forecasting model of hatching eggs is established by Mahalanobis Distance, which is with band regions range being 4 119.20～9 881.46 cm-1, principal component number being 19 and spectral pre-processing method being SNV+first derivative+Norris differential filter. The precision rate of calibration set is 92.5% and that of validation set is 91.67%. The study provides a new way for nondestructive testing of the fertile eggs and infertile eggs prior to incubation.