Research On Panoramic Super-resolution Localization Image Stitching Based On Localization Data

Super-resolution localization imaging(Single-molecule Localization Microscopy,SMLM)breaks the optical diffraction limit and brings optical microscopy to the nanoscale resolution,providing a new opportunity for biologists to better observe the fine structure of biological macromolecules.The purpose of researchers is to get panoramic super-resolution images of the entire sample to count quantity and capture rare events much better.However,the field of view collected at a single time in actual imaging is often much smaller than the entire biological samples.Therefore,a large number of acquired super-resolution images need to be stitched to form a panoramic super-resolution images.The existing super-resolution image stitching algorithm based on localization data can not calculate the registration offset of the region without structure or little structure,and can not effectively complete the panoramic image stitching.To solve the above problems,this paper proposes a panoramic super-resolution image stitching framework based on localization data,aiming at improving the stitching performance of panoramic super-resolution images.Firstly,in view of the existing super-resolution image stitching algorithm based on localization data,the mismatch of stitching is caused by the gap between cells and tissues.In this paper,a super-resolution localization data stitching framework PNoiseStitcher combined with background noise is proposed.The framework includes five steps: path planning,down-sampling,registration,de-duplication and fusion.In path planning,it is proposed to construct minimum spanning tree using the absolute value of the Euclidean distance,which can effectively reduce the cumulative error.In the registration process,a registration algorithm combining background noise and Gaussian mixture probability model is proposed to solve the problem that the Gaussian mixture probability model can not calculate the registration offset of the region without structural points.Finally,the framework is used for super-resolution monochromatic image stitching,and three experiments are designed to verify the effectiveness of the proposed stitching framework.The experiments prove that PNoiseStitcher has better stitching performance and can effectively complete the stitching of panoramic super-resolution images.Secondly,in order to solve the registration error problem between multiple channels of multi-color super-resolution image stitching,a multi-channel matching mechanism was introduced based on the monochromatic registration framework,and a multi-color super-resolution localization data image stitching framework was proposed.The framework firstly stitched the data of the localization data obtained by multiple channels,and then all the stitched monochrome panoramic super-resolution localization data were input into the multi-channel matching mechanism to obtain the final multi-color panoramic super-resolution images.The multi-channel matching mechanism uses the biological tissue characteristics of different channels to adjust the registration.Finally,the performance of the proposed framework is verified in the simulation and experimental data sets,and the multi-channel matching mechanism can effectively can effectively adjust for misalignment.Experimental results show that the proposed multi-color super-resolution image stitching framework is feasible and effective,and the stitching accuracy is better than other algorithms.Finally,according to the proposed panoramic super-resolution image stitching algorithm based on the localization data,this paper designs and implements a image stitching plug-in PNoiseStitcher based on the localization data.The plug-in is designed using JaveEE and Matlab framework,including two modules: monochrome splicing,two-color stitching and rendering module,which realizes the stitching and rendering of monochrome and two-color images and provides convenience for users to analyze the biological characteristics of cell tissues.The panoramic super-resolution image stitching framework based on localization data provides a convenient tool for researchers to obtain panoramic images of the whole sample,and lays a foundation for subsequent biological information processing and analysis,which has important research significance and application value.