Abstract:Modeling image sets or videos as linear subspaces is quite popular for classification problems in machine learning. However, affine subspace modeling has not been explored much. In this article, we address the image sets classification problem by modeling them as affine subspaces. Affine subspaces are linear subspaces shifted from origin by an offset. The collection of the same dimensional affine subspaces of $mathbb {R}^{D}$ is known as affine Grassmann manifold (AGM) or affine Grassmannian that is a smooth and noncompact manifold. The non-Euclidean geometry of AGM and the nonunique representation of an affine subspace in AGM make the classification task in AGM difficult. In this article, we propose a novel affine subspace-based kernel that maps the points in AGM to a finite-dimensional Hilbert space. For this, we embed the AGM in a higher dimensional Grassmann manifold (GM) by embedding the offset vector in the Stiefel coordinates. The projection distance between two points in AGM is the measure of similarity obtained by the kernel function. The obtained kernel-gram matrix is further diagonalized to generate low-dimensional features in the Euclidean space corresponding to the points in AGM. Distance-preserving constraint along with sparsity constraint is used for minimum residual error classification by keeping the locally Euclidean structure of AGM in mind. Experimentation performed over four data sets for gait, object, hand, and body gesture recognition shows promising results compared with state-of-the-art techniques.

Metamorphosis of Images in Reproducing Kernel Hilbert Spaces

Skeleton-driven 2D Distance Field Metamorphosis Using Intrinsic Shape Parameters

Automating Image Morphing Using Structural Similarity on a Halfway Domain

Stochastic Metamorphosis in Imaging Science

Image reconstruction through metamorphosis

The Euler-Poincare theory of Metamorphosis

Stochastic metamorphosis with template uncertainties

Selective metamorphosis for growth modelling with applications to landmarks

Space-time metamorphosis

Convergence of the Time Discrete Metamorphosis Model on Hadamard Manifolds

A deep residual learning implementation of Metamorphosis

Image Metamorphosis Based on Field Morphing and Mesh Warping

Technical Section: 2D shape morphing via automatic feature matching and hierarchical interpolation

Medical Image Reconstruction Using Kernel Based Methods

Diffeomorphic Measure Matching with Kernels for Generative Modeling

Composite Kernel Target Detection Based on Mathematical Morphology for Hyperspectral Imagery

Diffeomorphisms Groups and Pattern Matching in Image Analysis

Image Set Classification Using a Distance-Based Kernel Over Affine Grassmann Manifold

Sketching Image Morphing Using Moving Least Squares

Diffeomorphic Shape Matching by Operator Splitting in 3D Cardiology Imaging

An Operator-Splitting Approach for Variational Optimal Control Formulations for Diffeomorphic Shape Matching