Complex Sample Signal Analysis And Multi-dimensional Data Fusion

With the increasing industrialization worldwide,there is a growing need to fully observe the microstructural information of samples.X-ray speckle imaging is a multimodal imaging developed in recent decades,which can be divided into speckle-tracking and speckle-scanning modes according to different data acquisition methods.However,due to the time-consuming use of digital image cross correlation algorithms to match samples and reference subsets,optimizing the cross-correlation algorithm and using two imaging methods in laboratory sources to demodulate multimodal images of target sample for comparison is the first research focus of this article.In addition,speckle-tracking based X-ray microtomography allows the reconstruction of tomographic images using absorption and phase projections and complements the information with dark field images.Multidimensional multimodal imaging and data fusion analysis using synchrotron X-ray phase-contrast microimaging and Fourier transform infrared microspectroscopy to obtain complete structural information of complex samples is another important direction of research in this work.This study revolves around multi-dimensional and modal imaging and data fusion analysis,with the following work:1.Since the traditional scattered image demodulation method,digital image interpolation algorithm,is time-consuming due to window indexing,the Box filter fast correlation algorithm is introduced for optimization,using the intermediate calculation of the participation in the subsequent operation,which greatly reduces the operation time.Compared with the previous works,the demodulation time of each set of multimodal images is reduced by more than 15 minutes.2.Using a solid diamond/tungsten target source,we experimentally verified how the speckle size and visibility v affect the speckle pattern quality by varying the number of sandpaper layers and the object-image distance.Suitable sandpaper was selected for different samples,and two methods based on single exposure speckle tracking mode and one-dimensional speckle scanning mode were used to demodulate the speckle signals of multi-density standard samples and multistructured bamboo samples,and multimodal demodulation patterns including sample absorption,refraction and scattering signals were successfully obtained.The potential of X-ray speckle imaging for demodulation of complex samples was demonstrated.3.In the work of X-ray multi-dimensional and modal data fusion and analysis,firstly,the bimodal 3D visualization of bamboo samples was successfully achieved,and fine fiber and pore structures were obtained on the same position slices.In addition,phase projection was used to successfully reconstruct the refractive index real 3D information of fish bone,and comprehensive analysis was conducted in conjunction with dark-field images.Finally,a synchrotron-based X-ray phase contrast microscopic tomography combined with Fourier transform infrared microspectral technology was used to reveal the 3D microstructure characteristics and biomolecular distribution of lung tumors,demonstrating the application prospects of multi-dimensional data fusion diagnosis in the field of biomedical imaging.