Study On Multispectral Data And Crop Chlorophyll Content Correlation

With high resolution, strong spectral continuity and rich information, the spectral remote sensing technology can provide the enough information of ground subject in a dynamic, fast, accurate and timely way. And it can detect the physiological and chemical changes during the leaf growth of crop, and it can also effectively indicate the contents of leaf water, chlorophyll and microelement and then-changes by analyzing the spectral data, and the corresponding spectral response curve of the leaf of crop can be showed. Based on these characteristics of spectroscopy, we will know the nutritional status in the process of crop growth.Currently, the spectral technology research on chlorophyll content has achieved more findings, and researchers mainly use spectral reflectance to set up model and then make predictions by modeling analysis. With the corn leaf images taken with multi-spectral camera herein as a data source, we obtained three sets of data in the most important period of corn growth (seedling stage, vegetative growth stage, heading stage), and taking fifteen to twenty sample points from each image, we obtained gray values of eight brands (425nm～850nm) of each point on the image. And the chlorophyll of the leaves can be determined by using the SPAD meter in the meanwhile. Then we built a predictive model by using linear regression analysis and obtain the linear equations between each spectrum band and the responding chlorophyll content. We built the linear relationship between the gray values and the responding chlorophyll contents and then analyze the effect of different bands on the chlorophyll contents by the means of the principle component analysis so that the sensitive bands can be found. We found that the sensitive bands in seeding stage is 425nm,475nm,550nm,575nm,615nm,675nm, the sensitive bands in growth stage is 425nm,550nm,615nm,675nm, and the sensitive bands in heading stage is 550nm, 675nm. Finally, the data were divided into five classes according to the chlorophyll contents, and then the support vector machine (SVM) is trained by using some data and the other date are used to predict the chlorophyll contents of the corn in three different growth stages.