Research On The Optimization Of Planting Density Based On 3D Reconstruction For Soybean Planted By Pulan Seed Company

Under the background of unbalanced development between population explosion and crop productivity,the optimization of crop yield and planting density is the key point to improve crop yield.However,at present,most of the studies on the optimum planting density of crops adopt a single method.Moreover,the accuracy of the optimization results of crop optimal planting density by the model method is low.To solve this problem,on the premise of combining with the booming 3D reconstruction technology in the field of agriculture at present,this research proposes an optimization method of planting density for soybean planted by Pulan Seed Company based on the low-cost 3D reconstruction platform.This method realizes the 3D reconstruction of plants during the whole growth period which is based on a 3D sampling platform for plants,takes machine vision(image processing and 3D reconstruction)as the technical means,takes data mining as research method.Therefore,the optimization model of crop phenotype precision extraction and soybean yield-density is realized.Five varieties of DN251,DN252,DN253,HN48 and HN51 were selected as experimental materials.The field experiment was conducted at the soybean experimental base of Northeast Agricultural University.The results show that the Pearson correlation coefficients of each phenotype are higher than 0.98 in the analysis of artificial measurement values and model extraction values.Based on the phenotypic parameters extracted from the 3D reconstruction models,the "Phenotypic Fingerprint" drawing,Logistic growth model and the correlation analysis with the yield of soybean plant related phenotypes during the whole growth period are respectively carried out in this research.The "Phenotypic Fingerprint" can obtain the rules of phenotypic changes with internal links,which has become an important tool for crop selection.According to the Logistic growth model of soybean plants,the time points of maximum growth rate of five soybean varieties can be determined.And the time points of maximum growth rate can provide basis for water and fertilizer supply of crops.According to the correlation analysis between phenotypes and yield(seed weight per plant)in the whole growth period,the most relevant periods of plant height,plant length,plant width,canopy height,canopy area,and plant volume are about R5,R7,R7,R5,R7 and R7,respectively.This rule provides theoretical support for optimizing planting density.This provides theoretical basis and clear direction for improving the light condition of crop canopy,setting reasonable planting density and increasing crop yield.Based on the above analysis,the optimization model of soybean planting density and the prediction model of soybean yield are put forward.Through this model,the optimal planting densities of DN251,DN252,DN253,HN48 and HN51 in the same region are16,000 plants per mu,24,000 plants per mu,16,000 plants per mu,19,000 plants per mu and 18,000 plants per mu,respectively.This model links the 3D reconstruction technology with the optimization of planting density for the first time,which undoubtedly opens up a new way to realize the optimization management of agricultural refinement.The optimization of soybean planting density by this model can effectively reduce the experimental intensity of current planting density research field.The method proposed in this research not only improves the level of automation in agriculture,but also realizes the optimization and popularization of industrial engineering in the field of agriculture.