基于多光谱成像技术的水稻特征光谱提取

为了获取水稻光谱的有效特征信息,选取健康的TN1#水稻幼苗为研究对象,利用由液晶可调谐滤波器、单色CMOS相机与计算机控制软件组成的多光谱图像采集系统,获取健康水稻幼苗的20个可见光通道的多光谱图像。在此基础上,采用多光谱图像的平均灰度值,通过波段选择的指数方法计算出各通道的波段指数并加以排序,选出波段指数较大的10个通道,目的是探讨能有效反映出水稻特征光谱信息的特征波段。实验结果表明,用波段选择的指数方法提取多光谱图像的特征波段,能快速获取水稻的叶片信息。通道475nm、500nm、530nm、545nm、550nm、520nm、560nm、630nm、660nm、720nm能更好地反映出水稻特征光谱信息,可作为水稻的有效特征信息通道。

Extracting feature spectrum of rice based on multi-spectral imaging technology

In order to obtain the effective characteristics of rice spectrum, healthy TN1 # rice seedlings are selected for study by using liquid crystal tunable filter, a monochrome CMOS camera and computer control software to get 20 multi-spectral images about healthy rice seedlings in visible wavelengths. On this basis, the average gray value of the multi-spectral images is acquired by experiment, the value of band index for each channel is calculated, and all bands are sorted through the band index method, then, 10 channels are selected in all bands to explore the characteristic bands which can effectively reflect the characteristic spectral information of rice. The experimental results show that extracting the characteristic band of multispectral images can quickly obtain the information of rice leaves by way of band index. Bands 475 nm, 500 nm, 530 nm, 545 nm, 550 nm, 520 nm, 560 nm, 630 nm, 660 nm, 720 nm can reflect the characteristics of spectral information, and they can be used as effective characteristic channels of rice.