Random Projections on Manifolds of Symmetric Positive Definite Matrices for Image Classification
This work addresses the challenge of efficiently incorporating Riemannian geometry into image classification for computer vision applications, offering a novel but incremental improvement over existing embedding techniques.
The paper tackled the problem of using Symmetric Positive Definite (SPD) matrices for image classification by proposing a random projection method that preserves manifold structure while enabling direct use of Euclidean-based learning algorithms, resulting in improved performance on tasks like face recognition and texture classification compared to recent methods.
Recent advances suggest that encoding images through Symmetric Positive Definite (SPD) matrices and then interpreting such matrices as points on Riemannian manifolds can lead to increased classification performance. Taking into account manifold geometry is typically done via (1) embedding the manifolds in tangent spaces, or (2) embedding into Reproducing Kernel Hilbert Spaces (RKHS). While embedding into tangent spaces allows the use of existing Euclidean-based learning algorithms, manifold shape is only approximated which can cause loss of discriminatory information. The RKHS approach retains more of the manifold structure, but may require non-trivial effort to kernelise Euclidean-based learning algorithms. In contrast to the above approaches, in this paper we offer a novel solution that allows SPD matrices to be used with unmodified Euclidean-based learning algorithms, with the true manifold shape well-preserved. Specifically, we propose to project SPD matrices using a set of random projection hyperplanes over RKHS into a random projection space, which leads to representing each matrix as a vector of projection coefficients. Experiments on face recognition, person re-identification and texture classification show that the proposed approach outperforms several recent methods, such as Tensor Sparse Coding, Histogram Plus Epitome, Riemannian Locality Preserving Projection and Relational Divergence Classification.