Information Extraction Technology In Interference Hyperspectral Imaging

Interference imaging spectroscopy, which is also named as Fourier Transformimaging spectroscopy(FTS),has many advantages such as huge throughput, multiplechannels, high accuracy and fine resolution, etc., making it one of the most importantaspect in spectroscopy research. With the rapid developments of high spaceresolution and spectrum resolution, the challenge is taken out with such huge dataset.Thus research on the data processing of the interference imaging spectroscopy canbetter support the development of this type imaging spectroscopy.Data processing for interference imaging spectroscopy covers four main aspects:pre-processing of the interferogram, the recovery of the information in interferogram,the space visualization and the acquisition and transfer of the interference data.The problem of image quality assessments is involved in every aspects of thewhole processing. The quality assessment can be used to check the processingsolution and system, to guide the arithmetic. So far, there is no criterion forhyperspectral image quality assessment, For the quality assessment of spectrumimages, in this paper, we regarded that quality assessment should base on practicalspectrum application, and proposed the Energy of Image Structure Distortionevaluation criterion based on spectrum image application.According to the process of the interference data processing, the four mainaspects are further researched respectively in the paper.For the interferogram pre-processing, an interference imaging data probabilitystatistic model is presented, the Maximum A Posteriori (MAP) is utilized tode-noising the interferograms. And a phase correction method is proposed based onphase correlation in the context of zero optical path difference in sampledinterference data.For the information recovery, modern spectrum estimation is introduced.Modern spectrum estimation technologies no longer take the unrecorded data as zeros,overcome one of the biggest flaws in Fourier transform method, and the spectralresolution can be improved significantly.For the need of displaying spectrum images in space domain, an interferencespectrum image space-domain displaying method is proposed based on interferencespectrum domain weights. This method no longer needs to extract the spectrum information. It calculates a set of weights coefficients by the space domaindisplaying needs. Through a series of interference data and weights operation, it candisplay space domain images both in gray order and multicolor. This methodconsumes much less display time and can be adopted to display interference imagesin real time.For spectrum image acquisition and transfer, compression technology is apressing issue. In this paper, statistic feature of interference image spectrum datawas first analyzed. Utilizing the features of the spectrum data, a two dimensionsymmetry, classification, and prediction compression algorithm is proposed. Also toaddress the development of three dimensional transform, a three dimensionanisotropy wavelet transform compression algorithm is proposed.At last, due to the mass data acquired by spectrometer, a PCI-E based mass dataacquisition and storage system and an Ethernet-based mass data processing systemwere designed, the framework of the two systems was introduced and keytechnologies were analyzed.