Neurogeometry of neural functional architectures

The term “neurogeometry” denotes the geometry of the functional architecture of visual areas. The paper reviews some elements of the neurogeometry of the functional architecture of the first visual area V1 and explains why contact geometry, sub-Riemannian geometry, and noncommutative harmonic analysis are brought in as natural tools. It emphasizes the fact that these geometries are radically different from Riemannian geometries.     

Nonlocal patches based Gaussian mixture model for image inpainting

We consider the inpainting problem for noisy images. It is very challenge to suppress noise when image inpainting is processed. An image patches based nonlocal variational method is proposed to simultaneously inpainting and denoising in this paper. Our approach is developed on an assumption that the small image patches should be obeyed a distribution which can be described by a high dimension Gaussian Mixture Model. By a maximum a posteriori (MAP) estimation, we formulate a new regularization term according to the log-likelihood function of the mixture model. To optimize this regularization term efficiently, we adopt the idea of the Expectation Maximization (EM) algorithm. In which, the expectation step can give an adaptive weighting function which can be regarded as a nonlocal connections among pixels. Using this fact, we built a framework for non-local image inpainting under noise. Moreover, we mathematically prove the existence of minimizer for the proposed inpainting model. By using a splitting algorithm, the proposed model are able to realize image inpainting and denoising simultaneously. Numerical results show that the proposed method can produce impressive reconstructed results when the inpainting region is rather large.     

Variational image inpainting

Inpainting is an image interpolation problem with broad applications in image and vision analysis. Described in the current expository paper are our recent efforts in developing universal inpainting models based on the Bayesian and variational principles. Discussed in detail are several variational inpainting models built upon geometric image models, the associated Euler‐Lagrange PDEs and their geometric and dynamic interpretations, as well as effective computational approaches. Novel efforts are then made to further extend this systematic variational framework to the inpainting of oscillatory textures, interpolation of missing wavelet coefficients as in the wireless transmission of JPEG2000 images, as well as light‐adapted inpainting schemes motivated by Weber's law in visual perception. All these efforts lead to the conclusion that unlike many familiar image processors such as denoising, segmentation, and compression, the performance of a variational/Bayesian inpainting scheme much more crucially depends on whether the image prior model well resolves the spatial coupling (or geometric correlation) of image features. As a highlight, we show that the Besov image models appear to be less interesting for image inpainting in the wavelet domain, highly contrary to their significant roles in thresholding‐based denoising and compression. Thus geometry is the single most important keyword throughout this paper. © 2005 Wiley Periodicals, Inc.     

A generalization of Cahn-Hilliard inpainting for grayvalue images

The Cahn-Hilliard equation has its origin in material sciences and serves as a model for phase separation and phase coarsening in binary alloys. A new approach in the class of fourth order inpainting algorithms is inpainting of binary images using the Cahn-Hilliard equation. We will present a generalization of this fourth order approach for grayvalue images. This is realized by using subgradients of the total variation functional within the flow, which leads to structure inpainting with smooth curvature of level sets. We will present some numerical examples for this approach and analytic results concerning existence and convergence of solutions. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Wavelet frame based blind image inpainting

Image inpainting has been widely used in practice to repair damaged/missing pixels of given images. Most of the existing inpainting techniques require knowing beforehand where those damaged pixels are, either given as a priori or detected by some pre-processing. However, in certain applications, such information neither is available nor can be reliably pre-detected, e.g. removing random-valued impulse noise from images or removing certain scratches from archived photographs. This paper introduces a blind inpainting model to solve this type of problems, i.e., a model of simultaneously identifying and recovering damaged pixels of the given image. A tight frame based regularization approach is developed in this paper for such blind inpainting problems, and the resulted minimization problem is solved by the split Bregman algorithm first proposed by Goldstein and Osher (2009) [1]. The proposed blind inpainting method is applied to various challenging image restoration tasks, including recovering images that are blurry and damaged by scratches and removing image noise mixed with both Gaussian and random-valued impulse noise. The experiments show that our method is compared favorably against many available two-staged methods in these applications.     

Simultaneous cartoon and texture image inpainting using morphological component analysis (MCA)

This paper describes a novel inpainting algorithm that is capable of filling in holes in overlapping texture and cartoon image layers. This algorithm is a direct extension of a recently developed sparse-representation-based image decomposition method called MCA (morphological component analysis), designed for the separation of linearly combined texture and cartoon layers in a given image (see [J.-L. Starck, M. Elad, D.L. Donoho, Image decomposition via the combination of sparse representations and a variational approach, IEEE Trans. Image Process. (2004), in press] and [J.-L. Starck, M. Elad, D.L. Donoho, Redundant multiscale transforms and their application for morphological component analysis, Adv. Imag. Electron Phys. (2004) 132]). In this extension, missing pixels fit naturally into the separation framework, producing separate layers as a by-product of the inpainting process. As opposed to the inpainting system proposed by Bertalmio et al., where image decomposition and filling-in stages were separated as two blocks in an overall system, the new approach considers separation, hole-filling, and denoising as one unified task. We demonstrate the performance of the new approach via several examples.     

Sub-Lagrangians and sub-Hamiltonians on affine bundles

In this paper, sub-Lagrangians and sub-Hamiltonians are defined on anchored affine bundles as a natural extension of sub-Riemannian metrics. A duality considered between regular sub-Lagrangians and sub-Hamiltonians, gives the same solution of the Euler–Lagrange and Hamilton equations. Using the Pontryagin maximum principle, we prove that a similar situation of sub-Riemannian minimizers is encountered in this case, i.e., for a positive-definite sub-Lagrangian (sub-Hamiltonian), the locally arc-minimizing curves are either regular ones (as solutions of the Euler–Lagrange and Hamilton equations) or abnormal minimizers. Translated from Sovremennaya Matematika i Ee Prilozheniya (Contemporary Mathematics and Its Applications), Vol. 61, Optimal Control, 2008.     

On the Whitney extension property for continuously differentiable horizontal curves in sub-Riemannian manifolds

In this article we study the validity of the Whitney \(C^1\) extension property for horizontal curves in sub-Riemannian manifolds that satisfy a first-order Taylor expansion compatibility condition. We first consider the equiregular case, where we show that the extension property holds true whenever a suitable non-singularity property holds for the endpoint map on the Carnot groups obtained by nilpotent approximation. We then discuss the case of sub-Riemannian manifolds with singular points and we show that all step-2 manifolds satisfy the \(C^1\) extension property. We conclude by showing that the \(C^1\) extension property implies a Lusin-like approximation theorem for horizontal curves on sub-Riemannian manifolds.     

Quasiregular Mappings on Sub-Riemannian Manifolds

We study mappings on sub-Riemannian manifolds which are quasiregular with respect to the Carnot–Carathéodory distances and discuss several related notions. On H-type Carnot groups, quasiregular mappings have been introduced earlier using an analytic definition, but so far, a good working definition in the same spirit is not available in the setting of general sub-Riemannian manifolds. In the present paper we adopt therefore a metric rather than analytic viewpoint. As a first main result, we prove that the sub-Riemannian lens space admits nontrivial uniformly quasiregular (UQR) mappings, that is, quasiregular mappings with a uniform bound on the distortion of all the iterates. In doing so, we also obtain new examples of UQR maps on the standard sub-Riemannian spheres. The proof is based on a method for building conformal traps on sub-Riemannian spheres using quasiconformal flows, and an adaptation of this approach to quotients of spheres. One may then study the quasiregular semigroup generated by a UQR mapping. In the second part of the paper we follow Tukia to prove the existence of a measurable conformal structure which is invariant under such a semigroup. Here, the conformal structure is specified only on the horizontal distribution, and the pullback is defined using the Margulis–Mostow derivative (which generalizes the classical and Pansu derivatives).     

Hölder-topology of the Heisenberg group

Aequationes mathematicae - The Heisenberg group is an example of a sub-Riemannian manifold homeomorphic, but not bi-Lipschitz equivalent to the Euclidean space. Its metric is derived from curves...     

Constructing Invariant Fairness Measures for Surfaces

The paper proposes a rational method to derive fairness measures for surfaces. It works in cases where isophotes, reflection lines, planar intersection curves, or other curves are used to judge the fairness of the surface. The surface fairness measure is derived by demanding that all the given curves should be fair with respect to an appropriate curve fairness measure. The method is applied to the field of ship hull design where the curves are plane intersections. The method is extended to the case where one considers, not the fairness of one curve, but the fairness of a one parameter family of curves. Six basic third order invariants by which the fairing measures can be expressed are defined. Furthermore, the geometry of a plane intersection curve is studied, and the variation of the total, the normal, and the geodesic curvature and the geodesic torsion is determined.     

Cahn-Hilliard inpainting and the Willmore functional

The Cahn-Hilliard equation has its origin in material sciences and serves as a model for phase separation and phase coarsening in binary alloys. A new approach in the class of fourth order inpainting algorithms is inpainting of binary images using the Cahn-Hilliard equation. Like solutions of the Cahn-Hilliard equation converging to two main values during the phase separation process, the grayvalues inside the missing part of the image are oriented towards the binary states black and white. We present stability/instability results for solutions of the Cahn-Hilliard equation and their connection to the Willmore functional. In particular we will consider the Willmore functional as a quantity to find the optimal scale of the inpainting result. (© 2008 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim)     

Mass transportation on sub-Riemannian structures of rank two in dimension four

This paper is concerned with the study of the Monge optimal transport problem in sub-Riemannian manifolds where the cost is given by the square of the sub-Riemannian distance. Our aim is to extend previous results on existence and uniqueness of optimal transport maps to cases of sub-Riemannian structures which admit many singular minimizing geodesics. We treat here the case of sub-Riemannian structures of rank two in dimension four.     

A Coarea Formula for Smooth Contact Mappings of Carnot–Carathéodory Spaces

We prove the coarea formula for sufficiently smooth contact mappings of Carnot manifolds to Carnot–Carathéodory spaces. In particular, we investigate level surfaces of these mappings, and compare Riemannian and sub-Riemannian measures on them. Our main tool is the sharp asymptotic behavior of the Riemannian measure of the intersection of a tangent plane to a level surface and a sub-Riemannian ball. This calculation in particular implies that the sub-Riemannian measure of the set of characteristic points (i.e., the points at which the sub-Riemannian differential is degenerate) equals zero on almost every level set.     

On the sub-Riemannian geodesic flow for the Goursat distribution

Under consideration is the sub-Riemannian geodesic flow for the Goursat distribution. We find the level surfaces of the first integrals that are in involution and study the trajectories in the phase space whose projections to the horizontal plane are closed curves.     

Sub-Riemannian Curvature in Contact Geometry

We compare different notions of curvature on contact sub-Riemannian manifolds. In particular, we introduce canonical curvatures as the coefficients of the sub-Riemannian Jacobi equation. The main result is that all these coefficients are encoded in the asymptotic expansion of the horizontal derivatives of the sub-Riemannian distance. We explicitly compute their expressions in terms of the standard tensors of contact geometry. As an application of these results, we obtain a sub-Riemannian version of the Bonnet–Myers theorem that applies to any contact manifold.     

MRA contextual-recovery extension of smooth functions on manifolds

In a recent paper, the first author introduced an MRA (multi-resolution or multi-level approximation) approach to extend an earlier work of Chan and Shen on image inpainting, from isotropic diffusion to anisotropic diffusion and from bi-harmonic extension to multi-level lagged anisotropic diffusion extension. The objective of the present paper is to extend and generalize this work to nonstationary smooth function extension to meet the goal of inpainting missing image features, while matching the existing image content without apparent visual artifact. Our result is formulated as an MRA contextual-recovery extension for the completion of smooth functions on manifolds by deriving an error formula, from which sharp error estimates can be derived. A novel estimate for the biharmonic operator derived in this paper is a formulation of the error bound in terms the volume, as opposed to the diameter, of the image hole.     

Relation between Euler’s elasticae and sub-Riemannian geodesics on SE(2)

In this note we describe a relation between Euler’s elasticae and sub-Riemannian geodesics on SE(2). Analyzing the Hamiltonian system of the Pontryagin maximum principle, we show that these two curves coincide only in the case when they are segments of a straight line.     

The polynomial sub-Riemannian differentiability of some Hölder mappings of Carnot groups

The polynomial sub-Riemannian differentiability is established for the large classes of Hölder mappings in the sub-Riemannian sense, namely, the classes of smooth mappings, their graphs, and the graphs of Lipschitz mappings in the sub-Riemannian sense defined on nilpotent graded groups. We also describe some special bases that carry the sub-Riemannian structure of the preimage to the image.     

On geodesic equivalence of Riemannian metrics and sub-Riemannian metrics on distributions of corank 1

The present paper is devoted to the problem of (local) geodesic equivalence of Riemannian metrics and sub-Riemannian metrics on generic corank 1 distributions. Using the Pontryagin maximum principle, we treat Riemannian and sub-Riemannian cases in a unified way and obtain some algebraic necessary conditions for the geodesic equivalence of (sub-)Riemannian metrics. In this way, first we obtain a new elementary proof of the classical Levi-Civita theorem on the classification of all Riemannian geodesically equivalent metrics in a neighborhood of the so-called regular (stable) point w.r.t. these metrics. Second, we prove that sub-Riemannian metrics on contact distributions are geodesically equivalent iff they are constantly proportional. Then we describe all geodesically equivalent sub-Riemannian metrics on quasi-contact distributions. Finally, we give a classification of all pairs of geodesically equivalent Riemannian metrics on a surface that are proportional at an isolated point. This is the simplest case, which was not covered by Levi-Civita’s theorem. __________ Translated from Sovremennaya Matematika i Ee Prilozheniya (Contemporary Mathematics and Its Applications), Vol. 21, Geometric Problems in Control Theory, 2004.