Research On Hyperspectral Image Processing Method For Micro-Nano Structure Analysis

Hyperspectral Imaging(HSI)technology has developed rapidly since it was first used in remote sensing technology at the end of the last century.Today,HSI technology is not only used in remote sensing and earth observation,but also in agriculture,food processing,surveillance,eye health,environmental and chemical imaging and other fields.As an emerging technology,HSI technology combines advanced spectroscopy and imaging technology to provide sufficient information for spectroscopy and spatial analysis,so it is considered suitable for the study of distribution and characteristics of micro-nano materials.At present,the application of HSI technology has shifted from the detection of macroscopic objects to the detection of micro and nano materials,and it has shown great potential.The main goal of imaging micro-nano-scale materials using HSI technology is to visualize these small materials for classification,identification and analysis during the research process.This article mainly analyzes the micro-hyperspectral images of perovskite crystals(MAPb Br3)and a variety of fluorescent molecules in two micro-nano materials.Through the study of perovskite crystals,it is confirmed that the micro-hyperspectral image has a huge advantage in analyzing the characteristics of a single micro-nano material(such as thickness,structure,atomic composition ratio,etc.);by studying a variety of fluorescent molecules,it is confirmed that the microscopic hyperspectral image has great potential in the classification and identification of micro-nano materials.The research in this article is mainly to provide an auxiliary means to better evaluate the preparation of micro-nano materials.The main research contents and contributions are as follows:(1)Taking the perovskite crystal as an example,the image acquisition mode of the microscopic hyperspectral imaging system is verified,and the parameters in the image acquisition process are set reasonably to obtain high-quality images.Determine the final research mode by analyzing different image acquisition modes.This article mainly studies the transmission mode;(2)Analyzing the microscopic hyperspectral image of the perovskite crystal,by analyzing its characteristic spectrum curve and comparing it with the energy spectrum,it is found that the ratio of Pb and Br atoms of the MAPb Br3 crystal is different,and the characteristic spectrum curve will also be different.It was found that this difference was visualized by K-means clustering.This research can provide an auxiliary method for the preparation personnel to evaluate whether the crystal preparation is qualified,or it can help the preparation personnel determine the degree of decomposition of the crystal edge;(3)Analyze the microscopic hyperspectral images of various types of fluorescent molecules,and study the classification of various types of micro-nano particles.Useful wavelength bands are selected for research by analyzing the characteristic spectral curve of fluorescent molecules;because data collection requires an illumination system,there may be light source instability in the process of acquiring data.The original data is converted to absorbance to reduce light source instability.For experiments Although the filtered bands eliminate a lot of useless information,there is still redundancy between the data.The redundant information is eliminated through Principal Component Analysis(PCA)dimensionality reduction,and finally,different classification algorithms are compared.Analyze its classification effect on various types of fluorescent molecules.This study uses a more accurate and efficient way to determine the category of fluorescent molecules.