| According to the analysis of the National Plant Protection Professional Statistics2010-2020,the annual loss of rice due to diseases in China reaches 2.35% of the total production,and the yield loss due to diseases in serious areas is as high as 73.13%.Rice diseases,especially rice blast,known as "rice plague",occur on a large scale all year round,seriously threatening China’s food security,and rice diseases mainly rely on their conidia to spread,so accurate detection of rice diseases is of great significance.However,there are still some bottlenecks in airborne spore detection:(1)The air composition in the field field environment is complex,and there are some biotic microparticles(e.g.,pollen,other disease spores)and abiotic microparticles(e.g.,PM2.5,PM1),and the presence of these microparticles affects the detection and identification for rice disease spores.(2)The presence of microparticles with similar morphological dimensions in the air poses a significant interference to the detection.Therefore,this project proposes a microfluidic detection method for rice diseases based on multispectral diffraction image recognition technology.The work is carried out in the following aspects:(1)Designing an integrated microfluidic chip for separation and enrichment and parameter optimization.The study is based on the mechanism of microfluidic separation and purification of spores in air,physicochemical parameters of microparticles,fluid flow mechanism,and analysis of particle motion and force in microchannels.Optimize the chip structure,flow rate parameters and separation bend angle.(2)To study the mechanism of multispectral diffraction imaging of rice disease spores.We analyze the chemical composition and structural characteristics of disease spores,analyze the spectral absorption characteristics of spores in air,analyze the differences in the characteristics of different spores at different wavelengths through simulation and experiment,summarize and analyze the characteristic wavelengths of samples,construct characteristic parameters,and establish a multispectral diffraction identification model.(3)Designing multispectral diffraction fingerprint image to collect MEMS physical structures.Based on the principle of multispectral diffraction imaging,the diffraction imaging structure is designed,a reasonable image acquisition component is constructed,the light source module is configured,and the usability and accuracy of the multispectral diffraction recognition algorithm are experimentally verified.The experimental results show that the multispectral diffraction recognition device designed in this project can realize the recognition and classification of similar spores,and the recognition rates of rice blast spores and rice strain spores by this method are 98.4% and 98.7%,respectively.It was also able to clearly classify five samples in the classification experiment of five samples.The comparison with the other four detection methods showed that the detection method of this subject can well solve the problem of classification and identification of similar spores.In addition,the multispectral diffraction identification method has the features of rapid detection,low cost,small size,large field of view and high accuracy.The above results indicate that the method can be used for early detection,identification and classification of rice disease spores in complex environments. |
