| 摘要: |
| 【目的】柑橘黄龙病（Citrus Huanglongbing,HLB）是一种无法根治且易扩散的病害，建立柑橘黄龙病病情诊断及分类的方法，以及时发现并去除病株，防止感染其他果树。【方法】基于高光谱成像技术，利用最小噪声分离变换进行降维去噪、像元纯净指数获取纯净像元并建立训练集，通过Fisher判别法对柑橘黄龙病病情进行鉴别并分类。【结果】通过对训练集设置适当的门限值，柑橘黄龙病病情识别正确率达90%以上。【结论】利用高光谱技术进行柑橘黄龙病病情诊断具有较高的可行性。 |
| 关键词: 柑橘黄龙病 高光谱图像 MNF变换 Fisher判别法 PPI |
| DOI： |
| 分类号: |
| 基金项目:国家自然科学基金（青年科学基金）项目 （31201129）；现代农业产业技术体系建设专项(CARS-27)；广东省科技计划项目(2011B-020308009) |
|
| Identification and classification of citrus Huanglongbing disease based on hyperspectral imaging |
|
|
| Abstract: |
| 【Objective】This study aimed to establish an effective method to identify and classify Huang-longbing (HLB) disease so that diseased plants can be detected and removed timely.【Method】Based on hyperspectral imaging technology,this paper adopted Minimum Noise Fraction (MNF) transformation for dimensionality reduction and de-noising,created training set of pure pixels using Pixel Purity Index (PPI),and identified the level of citrus Huanglongbing disease with Fisher discriminant.【Result】Results showed that,the classification accuracy of citrus Huanglongbing disease based on hyperspectral imaging was >90% and the errorwas <20% by setting an appropriate threshold.【Conclusion】It is feasible to detect and classify Huanglongbing disease using hyperspectral imaging technology. |
| Key words: citrus Huanglongbing hyperspectral imaging MNF transformation Fisher discriminant PPI |
