Font Size: a A A

Research On Maize Plants Three-dimensional Reconstruction And Traits Extraction Based On Structure From Motion

Posted on:2021-03-22Degree:MasterType:Thesis
Country:ChinaCandidate:F R ZhouFull Text:PDF
GTID:2393330611983247Subject:Agricultural Electrification and Automation
Abstract/Summary:PDF Full Text Request
Maize is the most productive staple in the world.The phenotypic parameters of maize are closely related to its biomass and yield,so it is important to accurately measure the phenotypic parameters of maize.The traditional measurement of these traits is subjective,laborious and destructive.In this study,the autonomous crawler phenotyping robot developed in the earlier stage was used to collect images of maize plants from different perspectives outdoors(the collection interval was about 5~6cm).The maize plant model was reconstructed based on Sf M algorithm.Based on the algorithms of distance maximum traversal and triangulation,accurate and nondestructive measurements of plant height,stalk diameter and leaf area were achieved.The main contents and results are as follows:(1)Taking 80 pot maize plants at tasseling stage as the research object,the three-dimension model of maize plants was reconstructed based on Sf M,and traits such as plant height,stalk diameter,three dimension leaf area and two-dimension leaf area were extracted.Compared with the manually measured values,the determination coefficient(R~2)of plant height,stalk diameter and leaf area was 0.970,0.842 and 0.901,the mean absolute percentage error(MAPE)was 3.163%,4.760%and 19.102%,and the root mean square error(RMSE)was 3.557cm,1.540mm and 48.163cm~2,respectively.The maize plants were divided into low overground part biomass maize and high overground part biomass maize by the fresh weight of the overground part plant,meanwhile the plant trait such as height,minimum enveloping box volume of single plant,stalk diameter and the number of leaves were extracted with segmented point cloud data to calculate the P value by single factor analysis of variance.The measured P values were 0.0003,0.0004,0.3170 and0.2415,respectively,and the results proved that the traits of plant height and minimum enveloping box volume of single plant were able to distinguish the low overground part biomass maize and high overground part biomass maize evidently.(2)The 696 maize varieties in total 2748 plants potted maize were captured for reconstructing in the whole stages,in the experiment the external environment factors such as the natural wind,the camera’s focal length and the number of images were studied influenced the effect of maize three dimension reconstruction.The results showed when the natural force was breeze,the images was more,the focal length was F10,the effect of three dimension reconstruction is greater,which was content with actual reconstruction requires.The plant height of 25 maize varieties were extracted and analyzed dynamically in the whole growth period.(3)The 24 maize plants in the field were captured three times for reconstruction during the jointing stage and extracted plant height.Compared with the manually measured values,the determination coefficient of plant height was 0.907,the mean absolute percentage error was 6.480%and the root mean square error was 4.200cm.The result demonstrated that the system was also suitable for measurement of the filed maize.
Keywords/Search Tags:maize plant, three dimension phenotypic traits, structure from motion, three dimensional reconstruction, three dimension point cloud processing
PDF Full Text Request
Related items