基于直接正交信号校正的水稻冠层叶瘟光谱诊断

Discrimination of Rice Canopy Leaf Blast Based on Spectroscopic Techniques and Direct Orthogonal Signal Correction

采用直接正交信号校正-连续投影算法(DOSC-SPA)联用的组合处理方法,实现了应用可见/近红外光谱技术对水稻冠层叶瘟病的快速准确诊断。采集水稻冠层样本共120个,其中健康和染病样本各60个。用DOSC-SPA方法,对水稻冠层光谱数据进行直接正交化处理,然后通过连续投影算法提取有效波长(EW),建立了有效波长(775 nm)与叶瘟诊断判别的直接线性方程:Y=5.283X。应用该方程对预测集样本进行诊断判别,其判别准确率为95.0%,获得了满意的判别精度。结果表明,DOSC-SPA组合处理方法提取水稻冠层叶瘟诊断的特征波长是非常有效的,并且通过直接线性方程能获得满意的判别精度,为后续水稻大田叶瘟病情监测、喷药处理及相应病情监测仪器的开发提供了方法和依据。

Visible and near infrared （Vis/NIR） spectroscopy was investigated for the fast and accurate discrimination of rice canopy leaf blast using the combining processing of direct orthogonal signal correction-successive projections algorithm （DOSC-SPA）. 120 rice canopy samples were collected with 60 samples for healthy and blast rice canopy leaves,respectively. The calibration set consisted of 80 samples with 40 of healthy and 40 of blast. The remaining 40 samples were used for validation set. Using the new combination of direct orthogonal signal correction-successive projections algorithm,the spectral data of rice canopy was directly orthogonal processed,and then the successive projections algorithm （SPA） was applied for the selection of most important variables,named effective wavelengths （EWs）. Finally,a direct linear function was developed between the effective wavelength （775 nm） and rice canopy leaf blast for the discrimination rice canopy leaf blast,which was shown as Y=5.283X. The validation set was used to evaluate and assess the prediction performance of the developed direct linear function,and the discrimination ratio was 95%. An acceptable discrimination result and a parsimonious discrimination model were achieved by this direct linear function. The results indicated that the new combination of direct orthogonal signal correction and successive projections algorithm was a powerful way for the selection of effective wavelengths,and Vis/NIR spectroscopy was successfully applied for the discrimination of the rice canopy leaf blast,and the direct linear function achieved an acceptable discrimination ratio. This study provided a new approach for the further studies such as the on-field monitoring of rice blast,the treatment of spraying rice leaf blast,and the development of instrument for leaf blast monitoring.