The Research Of High-throughput Methodology For Crop Phenotyping Using Proximal Sensing Images

Crop phenotyping is quite complicated due to the interaction of genotype(G),environment(E)and management(M).The quantification of phenotyping parameters can be used in the exploration of the preponderant genotype selection,in order to realize the dynamic monitoring and quantitative analysis of the batch of breeding area,the high throughput phenotyping is the key technology.It is urgent to build the automated high-throughput platform for crop phenotypic information retrieval to meet the need of the efficient measurement technology of the precision agriculture in the future.Here we validate two different high-throughput phenotyping platforms to capture image sequence of crops under both field and greenhouse environment.Then the high-throughput phenotyping was developed based on the multi-source image data captured by different cameras to extract the parameters for different scales.The main contents are as follows:1.The parameters of wheat early vigour under the greenhouse environment were extracted based on the multi-views image method.An experiment was conducted in a glasshouse for two wheat genotypes which shows significant difference in the phenotyping during the seeding stage.The multi-view images were taken using a "vegetation stress" camera from emergence to the 6th leaf stage.Point clouds were extracted using Multi-View Stereo and Structure From Motion(MVS-SFM)algorithm,then the phenotypic parameters including tiller and leaf number,plant height,Haun index,phyllochron,leaf length,angle,and growth rate.There was good agreement between observed and estimated leaf length across both lines were calculated from reconstructed point cloud.Significant contrasts of phenotyping parameters were observed between the two lines and were consistent with manual measurements.2.The method of the image information interpretation can be used in the quantification of the parameters for crops in the field.Here we developed two different new image-based methods to estimate plot-level ground cover for three crop species(cotton,sorghum and sugarcane),from either an ortho-mosaic or undistorted images based on UAV system.Ground cover for individual plots was calculated by an efficient vegetation segmentation algorithm.There was a good agreement between the ground cover estimates from ortho-mosaic and undistorted images when the target plot was near the centre of image(cotton:R~2=0.97,sorghum R~2=0.98,sugarcane R~2=0.84).Although the undistorted images had a large range of ground cover estimates for the same plots,the reverse calculation provides a potentially precise method to estimate ground cover.3.The multispectral image data of whole wheat growth period can be used for the dynamic monitoring of the field scale normalized difference vegetation index(NDVI).Here we validate a high-throughput phenotyping platform to dynamically monitor NDVI during the growing season for the contrast wheat cultivars and managements.The images were rapidly captured by an unmanned aerial vehicle(UAV)carrying a multi-spectral camera(RedEdge)at low altitude.NDVIs for individual plots were extracted from the reflectance at Red and Near Infrared wavelengths represented in a reconstructed and segmented ortho-mosaic.NDVI measured by UAV and RedEdge camera were strongly correlated with those measured by hand held GreenSeeker(R~2=0.85)but were offset with UAV readings about 0.2 units higher.The high-throughput phenotyping platform captured the variation of NDVI among cultivars and treatments.During the growing season,the NDVI approached saturation around flowering time(-0.92),then gradually decreased until maturity.Strong correlations were found between image NDVI around flowering time and final yield(R~2=0.82).Given that the image NDVI includes signals from background(soil and senescenced leaves),ground cover from a high resolution hand-held camera was used to adjust the NDVI from UAV.This slightly increased the correlation between adjusted NDVI and yield.In this study,a variety of wheat phenotyping parameters were quantified based on the multi-view image method.The data mining algorithm developed for the plot scale in the field is suitable for high throughput phenotyping for applications in agronomy,physiology and breeding for different crop species and can be extended to provide pixel-level data from other types of cameras including thermal and multi-spectral models.Then the high-throughput phenotyping platform in this study can be used in agronomy,physiology and breeding to explore the complex interaction of genotype,environment and management.Data fusion from ground and aerial sampling improves the accuracy of low resolution data to integrate observations across multiple scales.The growth status of a variety of crop can be monitored accurately by the high-throughput phenotyping technology based on interpretation of the multi-source images both for the field and greenhouse environment,then realize the dynamic analysis of the variations of phenotyping parameters of the different crops.It provides a strong technical support for the selection of the prepondera,nt genotype in the breeding practice of agricultural production,and it is also a joint point of the multidisciplinary in developing and application of the precision and wisdom agriculture.