Segmentation For Stem And Leaves And Key Phenotype Traits Measurement Of Chinese Cabbage Sprout Based On Multi-View

Plant phenotype survey is an important basis for the selection and breeding of good varieties and gene function research in agriculture.It provides data support for plant growth pattern research and environmental role monitoring.At present,the methods of obtaining and investigating the phenotypes of leafy vegetables mainly rely on manual labor,which causes damage to plants and fails to achieve continuous growth monitoring,resulting in missing data for some traits.To address the above problems,this paper proposed a method for accurate extraction of key phenotypic parameters of Chinese cabbage sprout based on multi-view reconstruction and automatic stem and leaf segmentation.A high-throughput automatic phenotypic trait measurement software system was also developed to complete the growth status monitoring of different varieties of Chinese cabbage sprout.The main research contents and conclusions of this paper are as follows:(1)Research on 3D morphological reconstruction and organ segmentation algorithm of Chinese cabbage sproutAt present,the phenotypic data of Chinese cabbage sprout is mainly obtained by manual estimation,which is time-consuming and labor-intensive,and the existing plant phenotypic processing methods are poorly automated.In this paper,we proposed an efficient method for3 D morphological reconstruction and organ segmentation of Chinese cabbage sprout based on the advantages of computer vision technology,which has high precision and does not damage plants.The low-cost rotating device was used to acquire the sequence images of Chinese cabbage sprout and reconstruct them based on the multi-view geometry technique.Then,a pure3 D model of the plant with realistic scale was quickly established by point cloud preprocessing methods.On this basis,an improved automatic plant segmentation method based on super voxels was studied.The concavity criterion and machine learning index were used to achieve accurate segmentation and evaluation of stem and leaf organs of Chinese cabbage sprout.(2)Quantitative and accuracy assessment methods for key phenotypic traits of Chinese cabbage sproutThe data on key traits of Chinese cabbage sprout are missing,while automatic phenotypic measurement techniques for common crop plants are highly complex,limited in extracting traits and poorly scalable.In this paper,we proposed an improved method for automatic,nondestructive measurement and quantitative assessment of key plant phenotypic parameters based on morphological characteristics of Chinese cabbage sprout.Seven agronomic traits including plant height,plant width,leaf number,leaf length,leaf width,leaf area and plant color were accurately extracted.Firstly,quantitative traits were calculated based on principal component analysis and improved leaf vein point search algorithm.Then,the HSV model and mean color histogram were combined to measure the plant color traits in multi-view images,and the identification criteria for standard and nitrogen deficient planting of Chinese cabbage sprout were given.The experimental results showed that the proposed method had low overall error and strong ability to distinguish the color difference between plants,and was suitable for phenotypic analysis of leaf vegetables with irregular shape.(3)Study on the construction of a multi-temporal phenotype measurement system for Chinese cabbage sproutThe existing high-throughput plant phenotyping platforms are costly and complicated to operate,which are difficult to meet the practical needs of agricultural research units for dynamic monitoring of key phenotypes of Chinese cabbage sprout in multiple time sequences.In this paper,we designed and developed a simple and efficient software system for high-throughput phenotype measurement of plants based on Qt framework,and investigated the method of constructing the growth time series map of multi-species Chinese cabbage sprout.By integrating3 D phenotype processing and measurement techniques and providing parameter defaults for built-in algorithms,the phenotype investigation efficiency was improved.Furthermore,we used Origin to draw 3D waterfall diagrams to realize the monitoring of key quantitative trait trends of Chinese cabbage sprout during the whole reproductive period,providing effective technical means for accurate phenotype analysis and variety comparative evaluation.