论文部分内容阅读
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。
The direct and orthogonal signal correction - continuous projection algorithm (DOSC-SPA) combined with the combination of methods to achieve the application of visible / near infrared spectroscopy of rice canopy rapid and accurate diagnosis. A total of 120 canopy samples were collected, of which 60 were healthy and infected. Direct orthogonalization of rice canopy spectral data with DOSC-SPA method was performed. Then the effective wavelength (EW) was extracted by continuous projection algorithm. The direct linear equation of effective wavelength (775 nm) and leaf blast diagnosis was established: Y = 5.283X. The discriminant accuracy of the predictive set samples was determined using this equation. The discriminant accuracy was 95.0%, and satisfactory discrimination accuracy was obtained. The results showed that the combination of DOSC-SPA extraction of characteristic wavelengths of rice canopy leaf plague diagnosis is very effective, and through the direct linear equation can be satisfied with the accuracy of identification, rice leaf blight monitoring follow-up, spraying treatment and the corresponding The development of disease monitoring equipment provides the method and basis.