Adaptive Image Reconstruction Using Information Measures
The authors, motivated by problems in image processing, study the ill-posed problem of recovering a function \(u\) from the knowledge of the vector \(Au+B\nabla u\), where \[ A_iu:=\int_\Omega a_i u dx,\qquad B_iw:=\int_\Omega\langle b_i,w\rangle,\qquad i=1,\ldots,m. \] Given a regularizing functional of the form \[ I(u):=\int_\Omega h(u,\nabla u) dx, \] they study the following two questions: 1) The tolerance problem, i.e., the minimization of \(I\) subject to the constraints \[ \int_\Omega u dx=1,\qquad |Au+B\nabla u-c|<\varepsilon \] with the parameters \(c\), \(\varepsilon\) given; 2) The penalization problem, i.e., the minimization of \[ I(u)+\alpha|Au+B\nabla u -c|^2 \] subject to the constraint \(\int_\Omega u dx=1\). A typical choice of \(h\), in view of the applications, is \[ h(u,\xi):=\begin{cases} u\phi(\xi/u) & \text{if }u>0;\\ \phi^{\infty}(\xi) & \text{if }u=0;\\ +\infty &\text{otherwise}\end{cases} \] where \(\phi^\infty\) is the recession function of \(\phi\). In anisotropic models, one has also \(\phi(\xi)=\psi(|\xi|)\) with \(\psi\) convex. The condition \(\psi'(t)>>t\psi(t)\) leads (approximately) to the anisotropic smoothing which is considered by now a natural requirement in image analysis. The paper is mainly concerned with the existence theory for problems (1) and (2). The methods rely on duality theory rather than on compactness and lower semicontinuity of the energy functionals involved.
- Anisotropic \(\mathrm{BV}-L^{2}\) regularization of linear inverse ill-posed problems
- Image recovery via total variation minimization and related problems
- On a class of ill-posed minimization problems in image processing
- Regularization of ill-posed Mumford–Shah models with perimeter penalization
- A Variational Method in Image Recovery
- Self-adaptive image reconstruction inspired by insect compound eye mechanism
- Maximum entropy solution to ill-posed inverse problems with approximately known operator
- Posterior information-based image measurement matrix optimization
- Mathematics of electron tomography
- Approximate maximum entropy on the mean for instrumental variable regression
- scientific article; zbMATH DE number 1423114 (Why is no real title available?)
- Adaptive Reconstruction of Millimeter-Wave Radiometric Images
- A two-order time-fractional diffusion coupled system for image denoising based on the \(H^{- 1}\) norm
This page was built for publication: Adaptive Image Reconstruction Using Information Measures
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4507406)
