从大量的成像光谱数据中选择出有效的特征波段, 用于对园艺作物的特征进行快速识别和分类。基于LCTF(liquid crystal tunable filter)成像的优点在于能够连续改变光谱透过中心波长。首先实验选取园艺作物——萝卜叶片为研究对象, 获取萝卜叶片表面每隔5nm各波段图像的灰度信息；然后求出各波段的灰度值标准差和相关系数；最后通过波段指数排序选取萝卜叶片的有效特征波段。实验结果表明, 在530nm、550nm、535nm、555nm和715nm波段具有较理想的波段指数值, 这些波段离散度大, 信息量丰富且波段间相关系数小, 因此这些波段是识别萝卜叶片的有效特征波段。

In order to extract effective characteristic wavebands from a large number of imaging spectral data, they are used for rapid identification and classification about horticultural crops. The experiment selects the radish leaves as the research object so that obtaining gray information from the surface of radish leaves every 5nm wavelength, which is based on LCTF imaging having an advantage of changing spectral transmission center wavelength continuously; then, the standard deviation and correlation coefficient for the gray value of images are calculated; finally, the effective characteristic wavebands of radish leaves are extracted through the sorting of waveband index. The experimental result shows that there are five ideal waveband index values at 530nm, 550nm, 535nm, 555nm and 715nm respectively, these wavebands have much discrete degree, rich information and small correlation coefficient among every waveband. Therefore, these wavebands can be used as effective characteristic wavebands identification for radish leaves.