| Maize is one of the three staple foods in China and holds immense economic value.The global agricultural land area is diminishing each year,making it crucial to increase maize production on limited land to boost the yield per acre.However,in dense planting environments,the orientation and overlap of maize leaves have a substantial effect on light energy utilization,ultimately impacting crop growth and yield.Therefore,obtaining precise leaf orientation and overlap parameters is vital for accurate cultivation and precision planting of maize.This thesis focuses on the study of maize leaf orientation and analysis of overlap,taking maize plants in a large field environment as the research object.Point cloud data was collected using Lidar scanning technology,and after stitching and reconstructing the data,single maize plant point cloud data were cropped out.The principal component analysis algorithm was used to determine the leaf orientation of single maize plants.Through field experiments,the position of single maize stalks was calculated,and the reconstruction results were compared with real values to evaluate the accuracy of the algorithm.Combining the stalk position and plant orientation,the leaf overlap of neighboring plants was extracted,which provides a parametric index for quantifying the degree of dense maize planting.The main research results of the thesis are as follows:(1)Study on leaf orientation discrimination of maize plant:Point cloud data of individual maize plants were collected from multiple perspectives using a solid-state Li DAR device.The point cloud projection data of maize plants were linearly fitted using the least squares method,random sample consensus algorithm,and principal component analysis algorithm to obtain a growth orientation plane that can represent the leaf orientation of the plant.The orientation discrimination capability based on the least squares method and random sample consensus algorithm was found to be low,with determination coefficients(R~2)of 0.622 and 0.718 and RMSE of 23.41 degrees and 28.39 degrees,respectively,based on rotational experiments.The R~2 value based on the principal component analysis algorithm was 0.91,and the RMSE was 7.09 degrees.(2)Study on leaf orientation discrimination of maize plants in the field:Mechanical Li DAR was used to collect point cloud data of maize plant clusters in the field.Ground noise was removed using direct-pass filtering,and ROI region extraction was performed on the denoised point cloud data to obtain the point cloud data of plant clusters of interest.Statistical filtering was used to filter and denoise regional plant cluster data,and maize single plants in the field were cropped out using a manual interactive method.The orientation of single maize plant clusters was calculated using the principal component analysis method to discriminate the leaf orientation of maize plant clusters.The R~2 and RMSE values were 0.87 and 5.43 degrees,respectively,after comparison with real values.(3)Maize plant density measurement:The leaf overlap parameters were calculated based on the parameters of stalk position,plant spacing,and leaf orientation.The point density method was used to extract the stalk position and calculate the plant spacing from the point cloud data of single maize plants in the maize plants.The correlation coefficient and RMSE were 0.84 and 0.08 meter,after regression analysis with actual data.(4)Leaf orientation visualization interface design:A display interface for discriminating the leaf orientation of maize plants was designed using MATLAB App Designer.The interface allows interactive cropping of single plant data and clearly displays the information of plant leaf orientation,stalk position,and plant density,effectively improving the efficiency of discriminating the leaf orientation of maize plants in agricultural production. |
