Improvements in Blind Image Restoration

摘要: We present a revisal of blind image deconvolution technique for the restoration linearly degraded images, without explicit knowledge either original or psf- point spread function. Even scenes which consist finite support object over uniformly black, white grey background, this works fine. Occurrence includes certain types medical imaging, astronomical and (1-D) gamma ray spectra processing. The only information that is required are nonnegativity true size object. The procedure involves recursive filtering blurred to minimize convex cost new approach experimentally shown be more reliable have faster convergence than existing nonparametric ¯nite methods, situations in exact known. This thesis covers basic implementation NAS-RIF method, using steepest descent, followed by swarm optimization technique- ACO, optimize results.