Automatic Device And Methods For Life-Cycle 3D Phenotype Recognition Of Plant

Plant phenotypic parameters are an important basis for judging the pros and cons of varieties in the selection and breeding of crops.At present,plant phenotype measurement is mostly done manually,which is low in efficiency and large in error,and it is easy to cause damage to the plant itself.Due to the variability of plant morphology and overlap and occlusion,the phenotypic measurement based on two-dimensional image analysis is also difficult to reflect the complete plant phenotype information,which is easy to cause information distortion.Therefore,building a plant phenotype3 D recognition device and establishing a high-precision,automated,and non-destructive measurement method of plant phenotype throughout the growth cycle is an important basis for restoring the plant growth process,efficiently obtaining plant phenotypic parameters,and accelerating the selection and breeding of crop varieties.This article builds a whole growth cycle monitoring platform based on the working scenarios of agronomic researchers to realize the collection of environmental parameters and plant parameters in the process of plant breeding.Develop a full growth cycle phenotype 3D recognition automation device to realize the collection of full information in the plant selection and breeding process,realize an efficient and automated variety selection and breeding model,reduce the work intensity of agronomy researchers,and improve the efficiency of experiments.In this paper,research on plant three-dimensional reconstruction is carried out,and a high-precision plant three-dimensional point cloud reconstruction method suitable for equipment is established.The depth camera is used to collect multi-view point clouds,and after calibration,filtering,sampling,stitching and secondary filtering,nearly a hundred three-dimensional point cloud models of plants are merged from the multiview point clouds to provide a basis for subsequent research.This paper studies the problem of plant phenotypic parameter measurement,proposes a point cloud data extraction method,and constructs a series of non-destructive measurement methods of plant threedimensional phenotypic parameters through regression analysis with the true value: plant height,width,projected area,dynamic activity measurement Methods,plant width and leaf number measurement method based on leaf tip detection,DBSCAN-based plant leaf number measurement method,etc.,to achieve higher precision and faster extraction of plant geometric information.In addition,this paper studies the relationship between plant parameters and environmental factors,taking moisture and light as examples,using three-dimensional point cloud-based measurement methods as tools,and using cabbage and basil as experimental objects,starting from agronomic experimental methods.The effect of moisture and light on plant growth demonstrates the effectiveness and reliability of the equipment in agronomic research.To sum up,this article designed a plant phenotype 3D recognition device for the lack of plant three-dimensional reconstruction measurement equipment,the difficulty of obtaining high-precision three-dimensional models,and the difficulty of plant point cloud organ segmentation,to achieve rapid,accurate and complete plant phenotypes and other small plants Point cloud acquisition and splicing to realize the measurement of a variety of key plant parameters based on three-dimensional point clouds.The plant growth monitoring method and phenotypic parameter measurement method proposed in this paper are of great significance for the related agronomic research of crop variety selection and breeding.