基于光谱波段自相关的水稻信息提取波段选择  

Wavebands Selection for Rice Information Extraction Based on Spectral Bands Inter-Correlation

作　　者：王福民[1] 黄敬峰[1] 徐俊锋[1] 王秀珍[1] 

机构地区：[1]浙江大学农业遥感与信息技术应用研究所,浙江大学农业信息科学与技术中心,浙江省农业遥感与信息技术重点研究实验室,浙江杭州310029

出　　处：《光谱学与光谱分析》2008年第5期1098-1101,共4页Spectroscopy and Spectral Analysis

基　　金：国家自然科学基金项目(40571115);科技部科技基础性工作专项(2003DEA2C010-13)资助

摘　　要：通过大田试验，使用ASD光谱仪测量水稻不同生育期的冠层光谱，将光谱以10nm为步长进行合并，再将不同日期光谱的所有波段组合计算相关系数平方（R^2），生成R^2矩阵，并绘制R^2分布图。根据R^2越大，光谱波段之间冗余信息越多，R^2越小，水稻光谱波段信息含量越多的原则，在所有测量日期中选择出前100个R^2最小值对应的波段，将这些波段进行统计分析。结果表明，可见光区域各个波段之间和红外（近红外和短波红外）区域各个波段之间都含有大量冗余信息。水稻信息量丰富的波段主要集中在可见光的长波波段，红边波段，近红外第一和第二峰值波段，以及短波近红外第一峰前区（1530nm附近）和第二峰值区（2215nm附近）。比较水稻与其他植被对于最优波段的选择，400-410，630-650和1520-1540nm三个波段区间表现为水稻信息提取较为独特的波段。The hyperspectral remote sensing data usually involve hundreds or even thousands of narrow bands, which may be crucial for providing additional information with significant improvements over broad bands in quantifying biophysical and biochemical variables of agricultural crop. However, the huge data generated by hyperspectral systems, and the problems this presents for storage and analysis, have far prevented the routine use of such data. The objective of the present research was to identify the spectral bands in the visible and near-infrared range that were suitable for the study of rice. The hyperspectral reflectance of canopy in different development stages was measured in experimental field using a I nm-wide spectroradiometer but was aggregated to 10 nm-wide bandwidths to match the first spaceborne hyperspectral sensor, Hyperion. The correlation coefficients（r） between all the combinations of spectral bands were computed, and then they were converted to R^2 , which constituted R^2 matrices. The matrices were plotted against wavebands. The criterion of band selection is that the lower the R^2 value, the less the redundancy between two wavebands while the higher R^2 indicates that there is redundant information between two wavebands. According to the criterion, the wavebands corresponding to the first 100 minimum R^2 values were selected from all canopy spectra collected on different dates. And then these bands were analyzed. The results indicate that the visible and infrared （NIR and SWlR） themselves contain redundant information. The wavebands containing abundant information of rice are located in specific bands in the longer wavelength portion of the visible region, with secondary clusters in red edge region, in strongly reflective near-infrared region with relatively higher reflectance, in one particular section of short wave near-infrared （SWlR） （1 530 nm） and in the second maximum reflectance region of SWlR （2 215 nm）. Compared with the selected bands with other vegetation, rice seems

关 键 词：波段自相关 水稻 波段选择 

分 类 号：S127[农业科学—农业基础科学]