High-throughput analysis of maize azimuth and spacing from Lidar data  

作　　者：Zilong Wang Yiming Zhang Liqing Chen Delin Wu Yuwei Wang Lu Liu 

机构地区：[1]College of Engineering,Anhui Agricultural University,Hefei 230036,China [2]Anhui Province Engineering Laboratory of Intelligent Agricultural Machinery and Equipment,Hefei 230036,China

出　　处：《International Journal of Agricultural and Biological Engineering》2024年第5期105-111,F0002,共8页国际农业与生物工程学报（英文）

基　　金：supported by the Natural Science Research Project for Anhui Universities(Grant No.2023AH040140).

摘　　要：Efficient leaf azimuth angles and plant spacing are crucial for enhancing light interception efficiency in maize,thereby increasing yield per unit area.Traditional methods for measuring these traits are labor-intensive and prone to error.This study aimed to develop an accurate and efficient method for determining leaf azimuth angles and plant spacing in maize to improve understanding of field competition and support breeding programs.Utilizing light detection and ranging(Lidar)technology,3D point cloud data of maize plants were collected,enabling effective 3D morphological reconstruction through multi-frame stitching.Principal component analysis(PCA)was employed to determine the leaf azimuth angles of individual maize plants.Additionally,a method based on point density analysis was developed to identify the central axis position of single maize plants.Specifically,point density in the neighborhood of each point in the maize point cloud was calculated,with the central axis determined along the direction of highest point density.The integration of PCA-based leaf azimuth detection and point density analysis provided a robust framework for accurately determining leaf azimuth angles and plant spacing.In the detection of leaf azimuth angles,this method achieved an R2 of 0.87 and an RMSE of 5.19°.For plant spacing detection,the R2 was 0.83 and the RMSE was 0.08 m.This approach facilitates parameterized modeling of field competition,significantly enhancing the efficiency of breeding programs by providing detailed and precise phenotypic data.Despite the high accuracy demonstrated by the proposed methods,further investigation is needed to evaluate their effectiveness under varying environmental conditions and across different maize varieties.Additionally,challenges related to partial occlusions and complex canopy structures may impact the accuracy of point cloud data analysis,necessitating further refinement of the algorithms.

关 键 词：LIDAR maize azimuth angle 3D point cloud principal component analysis 

分 类 号：S513[农业科学—作物学]