| The detection of nitrogen in the leaves of kiwifruit tree during growth and development is of great significance for improving the yield and quality of kiwifruit.The traditional method of chemical test of nitrogen content in leaf of plants has disadvantages such as time-consuming and laborious,complicated process,detrimental to samples,and poor timeliness.It is difficult to achieve rapid and non-destructive nitrogen content in large-scale fruit trees in the production process in the current era for the requirements of real-time detection.In this paper,the kiwifruit leaves during the growth process were used as the research object based on the hyperspectral technique,a prediction regression model for the nitrogen content of kiwifruit leaves during the growth process.And it was established in the near-infrared region of 865.11-1715.03 nm.The rapid detection of nitrogen in the leaves of kiwifruit in the growth process were realized.The main research contents of this paper are as follows:(1)The paper determined the pretreatment method for spectral data of kiwifruit during growth.In the determination of nitrogen content in kiwifruit leaves,comparing the original spectral data without data preprocessing with different pretreatment methods such as FD,SD,MSC and SNV,the SNV method has better effect on the spectral data of kiwifruit leaves,and it has better correction ability to spectral data.(2)The feature extraction of the spectral data of kiwifruit leaves during the whole growth process and the prediction of nitrogen content of leaves were achieved.The PLSR and SVR were compared between the spectral data without characteristic processing and the spectral data after PCA,CARS extraction.And the SNV-CARS-SVR Rp was 0.8329 and the RMSEp was 0.1777,indicating that the SNV-CARS-SVR method can better predict the nitrogen content of the kiwifruit leaves throughout the growth process.(3)The prediction of spectral data extraction and nitrogen content of kiwifruit leaves during partial growth were realized,including two parts from May to July and from July to September.The PLSR and SVR were compared between the spectral data without characteristic processing and the spectral data after PCA,CARS extraction.The Rp of kiwifruit leaves from May to July using SNV-CARS-PLSR was 0.8507,and the RMSEp was 0.1289.In the July-September period,the Rp of the kiwifruit leaf using CARS-PLSR was 0.7478,and the RMSEp was 0.1839.The research shows that the rapid detection of nitrogen content in kiwifruit leaves during the growth process is achieved based on hyperspectral techniques,which provides a new method for the determination of nitrogen content in kiwifruit leaves.And it has broad application prospects in the application of kiwifruit industry. |