Spectrum Reconstruction And Imagign Stitchign Of A Temporally And Spatially Modulated Infrared Imaging Fourier Transform Spectrometer

The imaging spectrometry is the leading edge discipline of modern remote sensing and it is the combination of imaging technology and spectroscopy technology.The typical application of the imaging spectrometry is the imaging spectrometer,which can simultaneously acquire the two-dimension spatial information and one-dimension spectral information of the object by remote measurement.The imaging Fourier transform spectrometer is widely used in the mid-infrared.According to the way the optical path difference is produced,the imaging Fourier transform spectrometer can be classified into 3 types,temporally modulated,spatially modulated and temporally and spatially modulated.The temporally and spatially modulated imaging Fourier transform spectrometer has no moving mirror or slit,and thus has the advantage of multi-channel,high luminous flux and static interfering,which brings it the broad application prospects in environment monitoring,emergency,mineral exploration and atmosphere measurement.The spectral and spatial information can be used to find the type and location of the dangerous goods in emergencies.The data processing is the precondition of the imaging spectrometer application and the spectrum and image processing algorithm is different for different types of imaging Fourier transform spectrometer.This paper investigates the spectrum reconstruction and imaging stitching of a stepped mirror based temporally and spatially modulated imaging Fourier transform spectrometer with the three-dimension data cube.The spectrum of the target can be obtained by the spectrum reconstruction and the panorama can be obtained by the image stitching.The work mainly includes the following four parts:(1).The integration of the prototype is accomplished,and the three-dimension data cube is captured with it.(2).Since the stepped mirror is placed vertically and the light intensity in each sub-mirror are different,there should be vertical lines in the edge of the image corresponding to each sub-mirror.A polar Hough transform method is proposed to cutting the image in the data cube into image unit base on the characteristic of the image.(3).The spectrum reconstruction of the target is accomplished.The spectral information of the target,which is distributed in the data cube,is stitched in an image by directed image stitching method.The one-dimension interferogram is obtained by sampling the interferogram.A re-sampling method is proposed to deal with the inherent phase error caused by the sub-mirror height error.An empirical mode decomposition based method is proposed to enhance the spectral resolution of the instrument.The spectral resolution is enhanced by first decomposing the signal into different frequency parts and enhance the high frequency part while suppress the low frequency parts before reconstruction.(4).The image stitching is accomplished after feature point detection,feature description,feature matching and image fusion.To mitigate the light intensity change caused by the stepped mirror and retain the information of the image unit,the image is transformed to the HSI space to adjust its intensity component.