Research Of Stepwise Integrated Filter Based Infrared Spectral Imaging

As an important observation technology, space-borne spectral imagery has provided mass of high spatial/spectral resolution imaging data, contributing to the development of many industries. Still, the rapid development calls for higher spatial/spectral resolution and more spectral bands. However, the increasing spatial/spectral resolution results in much more difficulties constraining power, volume and weight in a spectral imager design. In this paper, key technologies of light Mid-Short wave infrared( MSWIR) spectral imager are researched from several aspects below, based on the recently rapid-developing stepwise integrated filter.According to the miniaturization of MSWIR spectral imaging and optimization of spectral splitting methods, different spectral splitting methods are compared. It reveals that dispersion-based splitting and filter-based splitting method are adapt to space-borne high frame rate spectral imaging scenario. The equivalence of Signal-to-Noise ratio(SNR) to assess the sensitivity of Short-wave infrared(SWIR) band and Mid-wave infrared(MWIR) band is proved. Then, adding with the atmospheric scattering and instrument background radiation, the traditional SNR model is modified. Based on the new model and system diffraction limit, a SWIR spectral imager orbiting Earth in a height of 500 km,is designed, it has a spectral resolution of 15 nm and ground sampling distance of 10 m. Design result shows that a stepwise integrated filter(IF) based spectral splitting method, with 6 TDI stages, could match a SNR of 100 across 2.0 μm ~2.5 μm, of which the aperture is close to the diffraction limit. The sensitivity it provides and the system volume it consumes is better than that of dispersion-based splitting methods.Based upon the projection process and the preliminary designed high precision IF-FPA match structure, a characteristic image pattern is determined. By comparing the least squares(LS) linear regression-based straight line edge sub-pixel localization method, area-based LS straight line edge sub-pixel localization method, and the Zernike orthogonal moment based straight line edge sub-pixel localization method, the optimal method, which is the last of the three, is selected to recognize the special image pattern. Thus the post assemble non-intrusive detecting process is completed. The experiment is carried out after the algorithm simulation. The result reveals that the IF has a 1.146901°rotational mismatch with the FPA, and the projected pass band width is 6.09 pixels. It means only rotational mismatch exists between the IF and FPA.A spectral calibration is carried out for the in-situ IF spectral imaging system. The calibration data proves that the spectral transmission curve in low temperature is just slightly different from that in room temperature. After that, a method to derive the optimal blocking band curve of the IF is proposed, of which the error is also given. Carrying it forward, a method to revert the optimal spectral response curve without the help of a monochromator is proposed, that is, with blackbody radiation in different temperature input to the system, a optimal set of parameters representing a guassian type spectral response curve could be found. And the optimal set of parameters is proved to be unique. An experiment is then carried out to test the conclusion. Besides, spectrum mismatch due to the pitch, roll, yaw of the platform is analyzed. Vice verse, the stability of the platform the spectrum matching process requires is also analyzed.In the end of this paper, a a SWIR IF spectral splitting based hyperspectral push-broom imaging experiment is presented. A 3D datacube is acquired, together with the false color image and the 64 bands of grey scale images. Some suggestions are proposed to increase the reversion efficiency: through parallel computing and cloud computing.