Design And Implementation Of System For Space Object Image Restoration And Feature Extraction

With the development of space technology, the national strategy for spaceis becoming more and more important. As an important part of the spacesecurity, space target recognition is indispensable. The research is on theimages which obtained by large optical telescopes. In the recognition process ofSpace target, there are two difficulties: First is the image blurring caused by theimpact of atmospheric turbulence; second is the influence of target posture andobservation angle to the target recognition. At the same time, with the need ofreal-time processing and analysis, selecting an appropriate algorithm is alsoimportant.Adaptive optics technology is the most effective measures to overcome theinfluence of atmospheric turbulence, but the adaptive optics compensation of theatmospheric turbulence is not completely. Details of the image are still seriousinhibition and attenuation. Therefore, to obtain a clearer adaptive optics imagesalso needs image post-processed. The generally algorithm of adaptive opticalimage restoration is complex and difficult to satisfy the needs of real-timeobservational. In the paper, it puts forward a kind of hardware based interactiveprocessing structure to solve the problem based on the image degradationmodel. The recovery process consist two steps, first is the estimation of pointspread function completed by computer, and second is the realization of thetraditional image restoration algorithm by embedded hardware system.The key is the extraction of affine invariant feature in space targetidentification. To meet the demands of rapid processing of large-capacity spatialdata, selecting an appropriate scheme of feature extraction and algorithmprocessing platform is essential. In this paper, to reduce the amount of dataprocessing, it choses the contour invariant feature of images. The choice is theboundary invariant moments and Fourier descriptors. The platform of featureextraction algorithm selected embedded hardware systems.According to the requirement of real-time and the algorithm flow for theadaptive optics image processing, we adopt a hybrid hardware configurationbased on FPGA and DSP as the design scheme. The real-time image processing board works stably and plays a good performance. It will do a help inthe observation and analysis of space targets.