FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2803580049> ?p ?o ?g. }

• W2803580049 abstract "Image restoration is a critical preprocessing step in computer vision,producing images with reduced noise, blur, and pixel defects.This enables precise higher-level reasoning as to the scene content inlater stages of the vision pipeline (e.g., object segmentation,detection, recognition, and tracking).Restoration techniques have found extensive usage in a broad range ofapplications from industry, medicine, astronomy, biology, andphotography.The recovery of high-grade results requires models of the imagedegradation process, giving rise to a class of often heavilyunderconstrained, inverse problems.A further challenge specific to the problem of blur removal is noiseamplification, which may cause strong distortion by ringing artifacts.This dissertation presents new insights and problem solving proceduresfor three areas of image restoration, namely (1) modelfoundations, (2) Bayesian inference for high-order Markovrandom fields (MRFs), and (3) blind image deblurring(deconvolution).As basic research on model foundations, we contribute to reconcilingthe perceived differences between probabilistic MRFs on the one hand,and deterministic variational models on the other.To do so, we restrict the variational functional to locally supported finiteelements (FE) and integrate over the domain.This yields a sum of terms depending locally on FE basis coefficients,and by identifying the latter with pixels, the terms resolve to MRFpotential functions.In contrast with previous literature, we place special emphasis on robustregularizers used commonly in contemporary computer vision.Moreover, we draw samples from the derived models to furtherdemonstrate the probabilistic connection.Another focal issue is a class of high-order Field of Experts MRFswhich are learned generatively from natural image data and yieldbest quantitative results under Bayesian estimation.This involves minimizing an integral expression, which has no closedform solution in general.However, the MRF class under study has Gaussian mixture potentials,permitting expansion by indicator variables as a technical measure.As approximate inference method, we study Gibbs sampling in thecontext of non-blind deblurring and obtain excellent results, yetat the cost of high computing effort.In reaction to this, we turn to the mean field algorithm, and showthat it scales quadratically in the clique size for a standardrestoration setting with linear degradation model.An empirical study of mean field over several restoration scenariosconfirms advantageous properties with regard to both image quality andcomputational runtime.This dissertation further examines the problem of blind deconvolution,beginning with localized blur from fast moving objects in thescene, or from camera defocus.Forgoing dedicated hardware or user labels, we rely only on the imageas input and introduce a latent variable model to explain thenon-uniform blur.The inference procedure estimates freely varying kernels and wedemonstrate its generality by extensive experiments.We further present a discriminative method for blind removal of camerashake.In particular, we interleave discriminative non-blind deconvolutionsteps with kernel estimation and leverage the error cancellationeffects of the Regression Tree Field model to attain a deblurringprocess with tightly linked sequential stages." @default.
• W2803580049 created "2018-06-01" @default.
• W2803580049 creator A5072831706 @default.
• W2803580049 date "2018-01-01" @default.
• W2803580049 modified "2023-09-23" @default.
• W2803580049 title "Foundations, Inference, and Deconvolution in Image Restoration" @default.
• W2803580049 hasPublicationYear "2018" @default.
• W2803580049 type Work @default.
• W2803580049 sameAs 2803580049 @default.
• W2803580049 citedByCount "0" @default.
• W2803580049 crossrefType "dissertation" @default.
• W2803580049 hasAuthorship W2803580049A5072831706 @default.
• W2803580049 hasConcept C106430172 @default.
• W2803580049 hasConcept C11413529 @default.
• W2803580049 hasConcept C115961682 @default.
• W2803580049 hasConcept C124504099 @default.
• W2803580049 hasConcept C154945302 @default.
• W2803580049 hasConcept C160633673 @default.
• W2803580049 hasConcept C174576160 @default.
• W2803580049 hasConcept C199360897 @default.
• W2803580049 hasConcept C2776214188 @default.
• W2803580049 hasConcept C2777693668 @default.
• W2803580049 hasConcept C2778045648 @default.
• W2803580049 hasConcept C30044814 @default.
• W2803580049 hasConcept C31972630 @default.
• W2803580049 hasConcept C41008148 @default.
• W2803580049 hasConcept C43521106 @default.
• W2803580049 hasConcept C49937458 @default.
• W2803580049 hasConcept C9417928 @default.
• W2803580049 hasConceptScore W2803580049C106430172 @default.
• W2803580049 hasConceptScore W2803580049C11413529 @default.
• W2803580049 hasConceptScore W2803580049C115961682 @default.
• W2803580049 hasConceptScore W2803580049C124504099 @default.
• W2803580049 hasConceptScore W2803580049C154945302 @default.
• W2803580049 hasConceptScore W2803580049C160633673 @default.
• W2803580049 hasConceptScore W2803580049C174576160 @default.
• W2803580049 hasConceptScore W2803580049C199360897 @default.
• W2803580049 hasConceptScore W2803580049C2776214188 @default.
• W2803580049 hasConceptScore W2803580049C2777693668 @default.
• W2803580049 hasConceptScore W2803580049C2778045648 @default.
• W2803580049 hasConceptScore W2803580049C30044814 @default.
• W2803580049 hasConceptScore W2803580049C31972630 @default.
• W2803580049 hasConceptScore W2803580049C41008148 @default.
• W2803580049 hasConceptScore W2803580049C43521106 @default.
• W2803580049 hasConceptScore W2803580049C49937458 @default.
• W2803580049 hasConceptScore W2803580049C9417928 @default.
• W2803580049 hasLocation W28035800491 @default.
• W2803580049 hasOpenAccess W2803580049 @default.
• W2803580049 hasPrimaryLocation W28035800491 @default.
• W2803580049 hasRelatedWork W1514073274 @default.
• W2803580049 hasRelatedWork W1542595032 @default.
• W2803580049 hasRelatedWork W2115374334 @default.
• W2803580049 hasRelatedWork W2140824929 @default.
• W2803580049 hasRelatedWork W2162596999 @default.
• W2803580049 hasRelatedWork W2188484476 @default.
• W2803580049 hasRelatedWork W2228350129 @default.
• W2803580049 hasRelatedWork W2330919853 @default.
• W2803580049 hasRelatedWork W2605145092 @default.
• W2803580049 hasRelatedWork W2790763184 @default.
• W2803580049 hasRelatedWork W2885442750 @default.
• W2803580049 hasRelatedWork W2949301738 @default.
• W2803580049 hasRelatedWork W3031135805 @default.
• W2803580049 hasRelatedWork W3043708496 @default.
• W2803580049 hasRelatedWork W3102025760 @default.
• W2803580049 hasRelatedWork W3123451359 @default.
• W2803580049 hasRelatedWork W3124839307 @default.
• W2803580049 hasRelatedWork W3128448487 @default.
• W2803580049 hasRelatedWork W2181930603 @default.
• W2803580049 hasRelatedWork W2468371437 @default.
• W2803580049 isParatext "false" @default.
• W2803580049 isRetracted "false" @default.
• W2803580049 magId "2803580049" @default.
• W2803580049 workType "dissertation" @default.