Research On Nutrition Information Of Facility Crops Growth And Development Of Mobile Detection Platform

Precise management of crop nutrients and moisture is an important part of precision agriculture, and accurate monitoring and diagnosis of crop nutrition is the premise and basis of precise management.Traditional analysis methods of chemical laboratory are time-consuming,timeliness bad and sampling and analysis process will cause damage to crops.Therefore,non-destructive testing technology on crop nutrition has been considered as very promising nutritional diagnostic techniques and has become a hot in current research of agricultural engineering.Based on facility lettuce and tomato as studyobjects,this paper propesed a non-destructive testing methodon crop nitrogen and water stress,using hyperspectral image,image technology and three-dimensional laser scan images with multi-sensor fusion information.On this basis,non-destructive testing system on facility crops growth information was developed. This paper completed following works:(1) Constructed an multi-sensor test information collection system.This paper studied breeding methodson crop nitrogen and water stresssample, using soilless cultivation technology to cultivatesample for traditional soil culture method is difficult to precisely controlcrop nutrition.It completed multi-sensor information collection of hyperspectralimages atdifferent levels of crop nutrients and water samples,image information and three-dimensional scanimages.(2) Features of hyperspectral images and characteristics of three-dimensional scanimageswith different nitrogen levels and water content of crop samples were studied.Based on hyper-spectral image data cube on different nitrogen levels in plant leaves and used stepwise regression sensitive areas combined with adaptive band selection method to extract characteristic spectral and features of imagesof crop nitrogen,and obtained meanintensity characteristic of featureimages.Characteristic compensation was done for errorcaused by crop nitrogen image features changed with moisture content and the crop nitrogen hyperspectral image diagnosis model was established;based on crop three-dimensional laser scanning data,using reverse engineering software Geomagic qualify to repair and smoothinterference noise and discontinuous intervals of three-dimensional data, and then obtained different nitrogen levels lettuce stem diameter,plant height,leaf area and biomass characteristicsthrough the establishment of spatial geometry crop 3D point cloud data.(3) Considering the limitation of single detection means in non-destructive testing of crop nuturition,this paper studiedcrop nitrogen,moisture content and multi-sensor-information technique and proposed to establish multi-scale fusion crop nutrient detection model with genetic algorithmcombined with PLSR,based on accurate extraction and analysis of cropshyper-spectral images, visual images and 3d laser scanning data.Results showed that the correlation coefficient R of the model is 0.95,the model precision is much better than single feature model with hyperspectral images,visual images and 3d laser scanning.The feature extraction algorithm and the eigenvectors in this study can provide reference for the development of online monitoring system for facility crop growing information.(4) The mobile NDT crop information systemis developed.The detection system can accomplish on-linenon-destructive testing and analysis forcrop nitrogen and moisture in the natural environment through multi-sensor information.