基于高光谱成像技术的番茄叶片灰霉病早期检测研究

摘要

The present paper put forward the technology route for feature images extraction of grey mold sick on tomato leaves based on SIMCA—combination image extraction based on MLR—grey mold sick information extraction based on minimum distance method.Firstly,through the 680～740 nm band's variance image and the discrimination power parameter,the feature band images was found,then the feature bands information was used as the input of MLR analysis,and under the 0.5 accuracy threshold value,99％ accuracy was obtained,which showed the discrimination power of the features bands for grey mold sick tormato leaf detection,and using the MLR regression coefficient to extract a band combination image,and through the minimum distance method,tomato grey mold sick information was found.The result shows that the proposed method has a very good prediction ability and greatly reduces the hyperspectral data processing time.%提出了独立软模式法(SIMCA)的番茄叶片灰霉病特征波段图像的提取,并通过多元线性回归法(MLR)提取波段融合图像,通过最小距离法获取番茄灰霉病患病信息的技术路线.利用680～740 nm波段的方差图像和建模能力参数提取的特征波段,并作为输入变量进行MLR分析,在0.5准确率阈值下,准确率均大于99％,说明特征波段可以实现番茄叶片灰霉病的检测,并利用MLR回归系数提取波段融合图像,通过最小距离法获取番茄灰霉病患病信息,结果表明所提出的方法具有很好的预测能力,为番茄灰霉病的早期检测提供了一种新方法,且大大降低了高光谱图像的数据处理时间.