大豆田间杂草的光谱识别研究

Study on Spectral Recognition of Weeds in Soybean Field

精确施药的关键是快速正确识别杂草。为此,利用ASD野外便携式光谱仪,在田间测量了大豆、马唐和稗草植株冠层在350～2 500nm波长内的光谱数据,经过数据预处理,数据分析波长选为350～1 300nm和1 400～1 800 nm。数据处理采用支持向量机(SVM)模式识别方法 ,用线性、多项式、径向基和多层感知核函数对大豆和杂草建立二分类模型。结果表明:三阶多项式核函数SVM分类模型的正确识别率最高,达到85%以上,且支持向量比例较小;以二分类模型为基础,利用投票机制建立了大豆、马唐和稗草的一对一多分类SVM模型,正确识别率达83%;田间光谱测量受光照、背景和仪器测量精度等条件的影响较大,但结果仍表明SVM结合光谱技术在田间杂草识别中的应用潜力很大。此研究为田间杂草识别及传感器的建立提供了研究思路和应用基础。

The key point of realizing precision chemical application is to correctly recognize weeds.A handheld FieldSpec R Spectroradiometer manufactured by ASD Incorporated Company in USA was used to measure the spectroscopic data of canopies of seedling corns,Dchinochloa crasgalli,and Echinochloa crusgalli weeds within 350~2500nm wavelength in the field.After date pre-processing,the effective wavelength range for spectral data process was selected as 350-1300nm and 1400-1800nm.Support Vector Machines(SVM) was selected as the method of pattern recognition to process data.Two-class classifier SVM models were built up respectively using 'linear','polynomial','RBF'(Radial Basis Function),and 'mlp(Multilayer Perception)' kernels.Comparison of different kernel functions for SVM shows that higher precision can obtained by using 'Polynomial' kernel function with 3 orders.The accuracy can be above 85%,but the SV ratio is relative lower.On the basis of two-class classification model,taking using of voting procedure,a model based on one-against-one-algorithm multi-class classification SVM was set up.The accuracy reaches 83%.Although the recognition accuracy of the model based on SVM algorithm is not above 90%,author still thinks that the research on weeds recognition using spectrum technology combining SVM method discussed in this paper is of tremendously significant.Because the data used in this study were measured over plant canopies outdoor in the field.Measurement is affected by illuminate intensity,soil background,atmosphere temperature and instrument accuracy.This method proposes a kind of research ideology and application foundation for weeds recognition in the field.