Research On Microscopic Hyperspectral Imaging System And Large Area Multidimensional Image Data Acquisition Method

Hyperspectral imaging technology originated in the field of multi-spectral remote sensing imaging in the 1970 s.With the ability of multidimensional information acquisition,it has been successfully applied for remote sensing detection.In recent years,this technology has been extended to the field of life science,especially the microscopic hyperspectral imaging technology combined with microscopy.This technology provides abundant spatial and spectral information for targets in pathological diagnosis and has promising application prospects.In this thesis,the intelligent acquisition method of micro-hyperspectral image and the corresponding image processing method are studied.Intelligent acquisition of micro-hyperspectral image and color image in the same field of view and reconstruction of large area data cube are realized by means of high precision stage control technology,microhyperspectral image auto-focus technology,multidimensional image mosaic technology and multidimensional image fusion technology,which increase the dimension of image data and lay the data fundamental for intelligent identification and analysis of biological tissue.In this thesis,the hardware and software design of the micro-hyperspectral imaging platform is given.Based on the imaging system,this thesis proposes the autofocus method for micro-hyperspectral image and the method for acquiring and generating large area multidimensional image data,including pre-processing method of micro-hyperspectral image,the screening method of blank micro-hyperspectral image,blank area detection method for micro-hyperspectral image,auto-focus method of the micro-hyperspectral image,the acquisition method and the generation method of the large area multidimensional image data of the micro-hyperspectral image.Aiming at the problems of auto-focusing of pathological micro-images,the existing image sharpness evaluation method and peak search method are improved to overcome the problem that the pathological micro-images have rough edges and weak texture to a certain extent and achieves accurate focus result of the image under single wavelength incident light.On this basis,the thesis utilizes the characteristics of wide spectral range of micro-hyperspectral image to compensate the position of the focal plane according to the spectral information and achieves the accurate focus result of the images under different wavelength incident lights.In addition,based on the single dimension image mosaic method,this thesis realizes the precise mosaic of micro-hyperspectral images by utilizing the spatial redundancy of micro-hyperspectral images and the location information of images.Experiment results show that the designed micro-hyperspectral imaging platform can automatically collect color and micro-hyperspectral image data in single field of view and multi-field of view.This platform relies on the designed storage platform and publishing platform to manage and publish image data,covering the whole life cycle of image data from acquisition,management to publication,which greatly improves the efficiency and automation of the collection of micro-hyperspectral image.With the support of large area multidimensional image data generation methods,the fusion data can provide more dimensional information reference for pathological tissue analysis.This method also offers a new idea for multidimensional image data fusion method,which has important practical significance and reference value.