This article reports on an investigation into the relationship of learners' use of metacognitive strategies to the EFL reading in test situation and non-test situation.The results suggested that(1) the use of meta...This article reports on an investigation into the relationship of learners' use of metacognitive strategies to the EFL reading in test situation and non-test situation.The results suggested that(1) the use of metacognitive strategies had a positive relationship to the reading test performance;and(2) successful test-takers reported significantly higher metacognitive strategy use than unsuccessful ones in the test situation and non-test situation;(3) When the same strategies were used,successful test-takers reported significantly higher metacognitive strategy use than unsuccessful ones in the test situation and non-test situation.展开更多
烟叶的化学成分是决定其香气、风味与吸食质量的关键因素,利用高光谱成像(hyperspectral Imaging, HSI)技术能够实现烟叶主要化学成分的快速、无损检测与可视化。本研究选取云南不同产区、不同等级的240份烟叶样本为研究对象,采集967.05...烟叶的化学成分是决定其香气、风味与吸食质量的关键因素,利用高光谱成像(hyperspectral Imaging, HSI)技术能够实现烟叶主要化学成分的快速、无损检测与可视化。本研究选取云南不同产区、不同等级的240份烟叶样本为研究对象,采集967.05~2561.33 nm范围内的高光谱图像,提出一种结合光谱差值与超像素聚类的感兴趣区域(region of Interest, ROI)提取方法,有效剔除背景、叶脉及不规则结构干扰。光谱数据经标准正态变量变换(standard normal variate, SNV)与一阶导数(first derivative, FD)联合处理后,采用主成分分析(principal component analysis, PCA)进行特征降维,并基于偏最小二乘回归(partial least squares regression, PLSR)对烟碱、总糖、还原糖、总氮、钾和氯6种成分进行统一建模。结果表明, SNV+FD预处理能有效提升模型性能,其中烟碱、总糖、还原糖和总氮的交叉验证决定系数(coefficient of determination for cross-validation, Q^(2))均超过0.89,交叉验证均方根误差(root mean square error of cross-validation, RMSECV)最低达0.09;在20个独立样本测试中, 6种成分的预测决定系数(coefficient of determination for prediction, R^(2))分别为0.930、0.908、0.854和0.915,平均相对偏差(relative difference, RD)小于5%,相对预测偏差(residual prediction deviation, RPD)均高于2.5,验证了模型的稳定性与预测能力。基于所建模型实现了烟叶主要化学成分的可视化分布,揭示了各成分在叶片不同区位的异质性特征。本研究提出的ROI提取与建模方法能够有效实现烟叶主要化学成分的无损检测与可视化,可为烟叶品质评价与精细化加工提供技术支撑。展开更多
文摘This article reports on an investigation into the relationship of learners' use of metacognitive strategies to the EFL reading in test situation and non-test situation.The results suggested that(1) the use of metacognitive strategies had a positive relationship to the reading test performance;and(2) successful test-takers reported significantly higher metacognitive strategy use than unsuccessful ones in the test situation and non-test situation;(3) When the same strategies were used,successful test-takers reported significantly higher metacognitive strategy use than unsuccessful ones in the test situation and non-test situation.
文摘烟叶的化学成分是决定其香气、风味与吸食质量的关键因素,利用高光谱成像(hyperspectral Imaging, HSI)技术能够实现烟叶主要化学成分的快速、无损检测与可视化。本研究选取云南不同产区、不同等级的240份烟叶样本为研究对象,采集967.05~2561.33 nm范围内的高光谱图像,提出一种结合光谱差值与超像素聚类的感兴趣区域(region of Interest, ROI)提取方法,有效剔除背景、叶脉及不规则结构干扰。光谱数据经标准正态变量变换(standard normal variate, SNV)与一阶导数(first derivative, FD)联合处理后,采用主成分分析(principal component analysis, PCA)进行特征降维,并基于偏最小二乘回归(partial least squares regression, PLSR)对烟碱、总糖、还原糖、总氮、钾和氯6种成分进行统一建模。结果表明, SNV+FD预处理能有效提升模型性能,其中烟碱、总糖、还原糖和总氮的交叉验证决定系数(coefficient of determination for cross-validation, Q^(2))均超过0.89,交叉验证均方根误差(root mean square error of cross-validation, RMSECV)最低达0.09;在20个独立样本测试中, 6种成分的预测决定系数(coefficient of determination for prediction, R^(2))分别为0.930、0.908、0.854和0.915,平均相对偏差(relative difference, RD)小于5%,相对预测偏差(residual prediction deviation, RPD)均高于2.5,验证了模型的稳定性与预测能力。基于所建模型实现了烟叶主要化学成分的可视化分布,揭示了各成分在叶片不同区位的异质性特征。本研究提出的ROI提取与建模方法能够有效实现烟叶主要化学成分的无损检测与可视化,可为烟叶品质评价与精细化加工提供技术支撑。
