Extraction Of Maize Planting Area And Growth Parameters Based On UAV Remote Sensing Data

Corn is one of the three most widely cultivated grains in the world and has been a major food crop in China for a long time.As one of the major crops in North China,corn accounts for about 1/3 of the total grain output in China.With the development of The Times,China has put forward the requirements of crop planting modernization,precision,and information development,which also put forward higher requirements for crop information extraction.Corn acreage and growth parameters can provide data support for the evaluation of corn growth and yield estimation and other agricultural affairs,and the statistics of corn acreage and growth evaluation are also an important prerequisite for macro-control of food prices,strengthening agricultural production management,and promoting the development of precision agriculture.Currently,the commonly used methods for plant information extraction include manual measurement with large labor consumption and low efficiency,satellite remote sensing with a wide range but poor data accuracy,and ground base station measurement with high accuracy but small range,which cannot meet the requirements of accurate extraction of crop information at field scale.To solve this problem,UAV remote sensing is used in this paper to study the accurate extraction method of field corn information.Image processing technology,point cloud processing technology,and deep learning techniques were used to accurately extract the planting area,plant height,and plant number of maize in the field.The specific research contents are as follows:（1）Data acquisition: The visible light data and lidar data of field corn were collected by UAV remote sensing platform,and the two kinds of data were solved by combining the base station coordinates to generate the visible light orthographic images and lidar point cloud data containing geographic coordinate information.（2）Planting area extraction: Firstly,the collected data are preprocessed,including point cloud data extraction,point cloud data denoising,multi-source data matching and clipping,and digital surface model generation;Secondly,an object extraction method based on feature segmentation is proposed.Vegetation index and local entropy feature are introduced to calculate the feature and segment the threshold of the data.Finally,the proposed method was evaluated not only on the collected field corn data but also on the standard data set to comprehensively verify the effectiveness of the proposed method.（3）Extraction of corn growth parameters: firstly,post-processing point cloud data,mainly including coordinate registration of two-stage point cloud data and point cloud coloring;Secondly,the field point cloud data were secured,and the row information of corn field was extracted by ground point filtering algorithm.The row data were removed and the row land was secured by the point cloud segmentation algorithm.Finally,the plant height and plant number of corn were extracted at the field scale.First,the coordinate set of the highest point of corn in the field was obtained by using the height seed point extraction algorithm based on the field height during the bare field period,and the plant height information of corn at the field scale was obtained.Secondly,the visible light data were processed by the Yolox network to obtain the number of corn plants at the field scale,the height change model was extracted by point cloud data,and the number of corn plants at the field scale was obtained according to the height change.Uav remote sensing has the advantages of flexibility,convenience and high precision of detection data.In this paper,the corn data obtained by UAV remote sensing is used to achieve the precise extraction of planting area and the extraction of plant number and plant height two growth parameters at the plot scale.