摘要
常用的整数阶一阶和二阶微分光谱曲线差异较大,而分数阶微分可实现光谱信息细化,能将整数阶微分容易忽略的信息充分利用起来,更能深度挖掘光谱中的潜在信息。以重金属铜胁迫的盆栽玉米为研究对象,在玉米叶片受不同浓度CuSO_4胁迫下,光谱反射率和Cu^(2+)含量的实验室测量数据基础上,对光谱反射率进行0~1内0.1阶的11个阶次微分处理,将各阶次的微分值与测得的玉米叶片中Cu^(2+)含量进行相关性分析;并对各阶次的微分光谱值进行0.01和0.05显著性水平的检验。分析结果表明,与常用的一阶微分相比,分数阶微分可以突显某些波段的光谱反射率与叶片中Cu^(2+)含量的相关性,扩大特征波段的选择空间。
The spectral curves processed by the integer first-order and second-order differentials have quite different,but the fractional-order differential can refine and fully use the spectral information which is easily ignored by differentials,and may better mining the spectral potential information in depth. The potted corn stressed by heavy metal copper was selected for the study,the spectra of corn leaves were processed by 0.1-order differential of 11 levels within 0~1 based on the measured data in laboratory that are the spectral reflectance values and Cu^2＋ contents of corn leaves stressed by CuSO_4 with different concentrations,and the correlation were analyzed between the differential values of each order and the Cu^2＋ contents in corn leaves,while the values of each order differential were tested in 0. 01 and 0. 05 significant levels. The analyzed results show that compared with first-order differential,the fractional-order differential can highlight the correlation between the spectral reflectance and the Cu^2＋ content in some bands,and expand the selection space of characteristic bands.
出处
《科学技术与工程》
北大核心
2017年第25期33-38,共6页
Science Technology and Engineering
基金
国家自然科学基金(41271436)
中央高校基本科研业务费专项资金(2009QD02)资助
关键词
高光谱遥感
盆栽玉米
分数阶微分
重金属铜
相关性分析
hyperspectral remote sensing potted com fractional-order differential heavy metal copper correlation analysis
