Bussgang Blind Deconvolution Algorithm And Application Research

In the research field of image processing, image restoration technology has alwaysbeen a difficult and hot problem. The purpose of image restoration is reconstructing theoriginal image from degraded image obtained from observation system. It is the basis ofimage processing, and have broad application prospect in medical imaging, remotesensing, astronomy, space detection and other fields. As an important branch of it, blindrestoration has become a research hotspot in recent years. In this dissertation, wepresent a unique multi-channel blind deconvolution framework derived from theBussgang algorithm which is applicable to multiple disciplines. This algorithm is namedas the self-correcting multi-channel Bussgang (SCMB) blind deconvolution algorithm,and it is able to solve the problems of serious optical image degradation and inversesynthetic aperture radar (ISAR) image defocusing by using of all available informationeffectively.Firstly, this paper introduces the basic principle of image blind restoration and theBussgang blind deconvolution and blind equalization technology is analyzed, then itpresents the gaussian mixture model parameters estimation method based on EMalgorithm. After that, this paper introduces several different types of point spreadfunction, and demonstrates the3-dimensional spatial imformation of three commonpoint spread functions by computer simulation.Then, this thesis introduces the basic theory of blind deconvolution by self-tuningmulti-channel Bussgang (SCMB) algorithm, and deduces the equalizing filter updatingalgorithm of Bussgang SCMB algorithm in detail. It explains that according to differentapplication, the choice of several nonlinear functions, the design principles of theprobability density function in gaussian mixture model and the feedback mechanism.Through comparison of the three different types of fuzzy image blind restorationsimulation results, it proved that the SCMB blind deconvolution algorithm is feasibleand reasonable, and it can achieve the purpose of image blind restoration with unknownblur types.Finally, the SCMB blind deconvolution algorithm is applied in self-focusing ofISAR image. The simulation experiment practical measured data shows that the SCMBblind deconvolution algorithm is feasible and reasonable, and it can achieve the aim ofISAR autofocus. Moreover, the versatility of this algorithm is confirmed by binaryimage blind restoration test and ISAR self-focusing experiment.