基于可见/近红外光谱技术的毛脚茧的光谱快速鉴别方法研究

基于可见/近红外光谱技术，研究了鲜茧毛脚茧快速鉴别的方法。采用主成分分析法对毛脚茧和成熟茧进行定性分析，再对光谱数据采用最小二乘支持向量机方法建模鉴别，通过连续投影算法从400-1000 nm波段的鲜茧光谱数据中选取了12个特征波长。采用最小二乘支持向量机法基于400 nm，430 nm，487 nm，512 nm，604 nm，616 nm，732 nm，759 nm，784 nm，852 nm，985 nm，999 nm这12个特征波长建模，结果对毛脚茧的判别准确率达到100%。表明可见/近红外光谱可以用于毛脚茧的鉴别检测。

Detection of Immature Cocoon Based on Visible and Near Infrared Spectroscopy

The visible and near infrared spectroscopy has been used to detect immature cocoon. Using principal component analysis of the qualitative analysis of the immature cocoon and mature cocoon, and then the spectral data using least squares support vector machine modeling identification method, through the successive projections algorithm 12 characteristic wavelengths selected from fresh cocoon spectral data of 400-1000 nm. By using the least squaresupport vector machine method based on 400 nm, 430 nm, 487 nm, 512 nm, 604 nm, 616 nm, 732 nm, 759 nm,784 nm, 852 nm, 985 nm, 999 nm of the 12 characteristic wavelength modeling, results of immature cocoon discriminate accuracy of 100%. Results show that the visible/near infrared spectroscopy can be used to identify the detection immature cocoon.