欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
为研究高光谱图像技术用于水稻种子活力快速无损鉴别的可行性,本试验以不同老化程度的4个水稻品种共960粒水稻种子为材料,对样品进行人工老化后进行发芽试验,统计发芽率和根长,计算简易活力指数,据此将每个品种的样品划分为不同活力梯度组,采用高光谱图像技术,通过提取水稻种子的光谱反射率,结合Savitzky-Golay（SG）平滑算法、标准正态变量（SNV）和多元散射校正（MSC）对874~1 740nm波段内的光谱数据进行去除噪声处理,采用主成分分析法（PCA）、连续投影算法（SPA）进行特征波长选择,基于全波段光谱和基于特征波长分别建立了偏最小二乘判别分析（PLS-DA）模型。试验结果表明,经MSC预处理后,采用SPA挑选的特征波长建立的PLS-DA模型,建模集和预测集的识别正确率分别达到100%和98.75%。研究结果表明,利用高光谱图像技术对水稻种子活力进行快速无损检测是可行的。
英文摘要：
In order to identify common rice seed vigor on the market rapidly and nondestructively,four kinds rice seed vigor have been identified by combining hyperspectral imaging technology and different discriminant models.The reflectance spectral were extracted from the region of interest in the sample images for analysis,the wavelengths from 874 to 1 740 nm were preprocessed by Savitsky-Golay（SG）,Standard Normal Variat（SNV）and Multiplicative Scatter Correction（MSC）.Partial Least Square-Discriminant Analysis（PLS-DA）were used to build discriminant models based on the preprocessed full spectra and selected sensitive wavelength by Principal Component Analysis（PCA）and Successive Projections Algorithm（SPA）from the preprocessed spectra.Among the discriminant models using the preprocessed full spectra and selected sensitive wavelength,PLS-DA models obtained the highest classification accuracy.The selected sensitive wavelength by SPA from the MSC preprocessed spectra,PLS-DA models obtained the best classification accuracy with 100%accuracy in both the calibration set and the predicated set.The results showed that it was feasible to identify rice seed vigor rapidly and nondestructively by hyperspectral imaging technology.
同期刊论文项目
同项目期刊论文
期刊信息
位置：