Research On Detection Method Of Single Maize Seed Vigor Based On Fluorescence Imaging Technology

Seeds are the most fundamental productive resource in agriculture,the most critical and essential part of the life cycle of various crops,and the core of various technological measures in agricultural science.The quality of seeds can have a significant impact on the yield of a crop and is often described in terms of seed vigor.Efficient detection of maize seed vigor facilitates the effective screening out of seeds that lack the vigor to achieve increased maize seed yield and thus increase maize production and contribute to the steady development of food production.However,traditional seed vigor detection methods have the disadvantages of long cycle time and high cost.Therefore,it is important to explore the microscopic mechanism and quality change characteristics in the process of seed vigor decline,to find the causes of seed vigor decline in maize from a microscopic perspective,and to achieve rapid detection of single seed vigor in line with the requirements of modern agricultural development.In this study,single-seed maize seeds were used as the research object,and the microscopic mechanism changes and vigor grade detection of the single-seed maize seed vigor decline process were studied based on microscopic fluorescence imaging technology and fluorescence hyperspectral technology.The main research contents are as follows:(1)Based on the micro-fluorescence imaging technique,combined with fluorescent probes,the changes in starch content and cell membrane structure during the aging of maize seeds were studied.Microscopic fluorescence spectra in the spectral range of 415-725 nm were used,pre-processed using SNV and SG convolution smoothing,and CARS was used as a feature band screening method.The screened feature bands can well reflect the substances associated with seed vigor and are consistent with the findings of fluorescence hyperspectroscopy.Based on the KNN algorithm as the classification detection model and combined with the conclusion of seed vigor detection experiments,seed vigor was classified into three categories: high vigor,low vigor,and no vigor.The optimized seed vigor classification detection model achieved an accuracy of 91%.High-accuracy classification of seed vigor was achieved based on microscopic fluorescence technology.(2)Based on the fluorescence hyperspectral imaging technology of 400-1000 nm,we carried out the research on the non-destructive detection method of maize seed vigor class.A fluorescence hyperspectral device was built for seed vigor detection.A 150 W UV light source was selected as the excitation light,and a surface array industrial camera was used to acquire the seed hyperspectral images,and a movable carrier table was equipped with self-adjustable speed.Different pre-processing methods,four different feature band screening methods,and three different machine learning algorithms were selected to model the acquired maize seed hyperspectral data;a further linear combination of texture data from fluorescence hyperspectral images was supplemented.The results show that the RAW-VCPA-IRIV-Feature-LDA model achieves the best classification results with a high accuracy of 93%.Based on fluorescence hyperspectral technology,high accuracy classification of maize seed vigor was achieved.In summary,this paper investigated the microscopic mechanism of the vigor decline process and vigor classification detection of single maize seeds using microfluorescence imaging technology and fluorescence hyperspectral technology.It laid the foundation for the development of maize seed vigor detection equipment.