DEBLURRING OF IMAGES USING BLIND SCHEMES

摘要: The thesis presents two blind deconvolution schemes for image blur removal. major types of has been worked out, namely, the gaussian and motion blur. corrupted by is reconstructed Evolutionary algorithm using pseudo-wigner distribution. second method deals with heuristically estimating parameter undergone effect mostly observed in astronomical imaging. deblurring blurred required due to hardware incapability capturing exact information moving object or camera. In this thesis, an assumed be dimensional convolution true a linear-shift invariant blur, known as point-spread function, psf, additive noise zero. implemented remove atmospheric turbulence modelled psf. proceeds randomly generating psf’s at each generation. generation are used estimate image. best fitted images then given input next After few generation, most feasible chosen. These closer estimated fused distribution reconstruct final inherent dynamic characteristic nature gives rise Whenever there relative between captured imaging system, that instant suered type A new heuristic approach framed out purpose parameter. This characterised its length direction. parameters restore direction from fourier domain iteratively computed Entropy RMSE quality metrics.