The Acquisition Research Of Rape Information Based On Hyperspectral Imaging Technology

Precision agriculture is modern agriculture which is based on information and knowledge,it can not only utilizate agricultural resources reasonably,increase crop yield and quality,reduce cost and improve the ecological environment,and then realize the sustainable development of agriculture, but also have a significant effect of promoting agricultural modernization the development of industries related to agriculture system.Quick and effective collection and describe crop growth information is an important basis of precision agriculture practice.According to the situation at home and abroad, this paper choose the Rape as the research object,uses hyperspectral imaging technology to determine the characteristic bands,and then to identify different kinds of rapeseed varieties.we studied the relationships of hyperspectral reflectance of rape leaves and chlorophyll content (the SPAD value).we also research how to identify different rapeseed varieties in different growth periods.This paper mainly studies the conclusion is as follows:(1)Applying hyperspectral image technology to identify different rapeseed varieties.Through principal component analysis of the hyperspectral data,it realized dimensionality reduction,and three characteristic bands were found out.The texture parameters were extracted from the optimal bands images based on gray level histogram and gray level co-occurrence matrix (GLCM) statistical methods. Artificial neural network model was used for the identification of varieties of rapeseed. Detection results of ANN model showed that calibration and prediction sets of rapeseed varieties were 93.75% and 91.67%, respectively.(2)The relationship between the reflectivity and chlorophyll content of rape leaves was studied, and establishing the quantitative analysis model of rape chlorophyll content. ANN was applied to establish the model between the spectral reflection values and SPAD values. The prediction results were obtained for the nitrogen content of rape leaf with the correlation of prediction of R=0.9237.(3)dentifying different rapeseed varieties in different growth periods, finding out the characteristic bands of rape leaves. obtaining features bands of the rape which are 544-565,674-693 and708-723nm, and using features bands to establish the identification models which have good results on the identification.(4)Studying the extraction methods of hyperspectral image information: hyperspectral data processing method was researched and the model of quantitative analysis was established.It shows that the hyperspectral imaging technology has a fast, nondestructive and accurate classification and identification effects on rape varieties.