Integration Of The Multi-spectral Camera System And Focusing Method Research

The images obtained by traditional RGB cameras have the same color fordifferent spectral. However, multi-spectral cameras don’t have this problem, for itpictures the target of the same position with several spectrum at the same time. So, thepictures obtained by multi-spectral cameras have more imaging features, are moreaccurate to restore the surface spectral emissivity. Multi-spectral cameras are widelyused in environmental monitoring, aerial photography, etc.This article is derived from the project of the Chinese academy of sciencesinstitute of remote sensing and digital earth. The project’s name is Wide-angleInfrared Dual-mode line/area Array Scanner, referred to as WIDAS for short. Theproject is composited by two multi-spectral camera and a thermal infrared thermalimager, can obtain visible light and near infrared wave band5five angles and thermalinfrared band7Angle datas.This article take the multi-spectral camera system from WIDAS system, tocomplete two aspects work mainly. First part is the integration of the multi-spectralcamera system, to design a DC/DC power module for the integrated system. Themulti-spectral camera need24VDC power when flying in the small plane. But therewas no suitable adapter, so the power module which converting28VDC power into24VDC power can be used. The second part is the imaging software of the camerasystem, using development kit of EPIX acquisition card and MFC interface to realizethe collection of the camera image data, five band image display functions.The research of multi-spectral camera focusing methods, use manual focus modedue to the multi-spectral camera don’t have automatic focusing mechanism. Theimage clarity evaluation can be a standard for advantages and disadvantages of thecamera focusing methods. For the image fuzzy because of incorrect focus, this paper proposes a no reference structure similarity algorithm, according to the presentcommonly used algorithm,such as mean square error, peak signal to noise ratio, thestructure similarity algorithm and so on.Then using Matlab R2009a software toimplemente these algorithms, and using the BP neural network model to establishmodel and test samples.Finally, the whole integration system is tested well, meet user’s basicrequirements, Power supply module’s efficiency was92%, within the error range. Noreference structure similarity algorithm can judge98%blurred images. It proved thefeasibility of this algorithm for blurred images away from the correct focus.