On the signature of an image

Abstract

Over the past decade, the importance of the 1D signature which can be seen as a functional defined along a path, has been pivotal in both path-wise stochastic calculus and the analysis of time series data. By considering an image as a two-parameter function that takes values in a $d$-dimensional space, we introduce an extension of the path signature to images. We address numerous challenges associated with this extension and demonstrate that the 2D signature satisfies a version of Chen's relation in addition to a shuffle-type product. Furthermore, we show that specific variations of the 2D signature can be recursively defined, thereby satisfying an integral-type equation. We analyze the properties of the proposed signature, such as continuity, invariance to stretching, translation and rotation of the underlying image. Additionally, we establish that the proposed 2D signature over an image satisfies a universal approximation property.

arXiv comment33 pages, comments welcome!

References in corpus (13)

• Rigid-Motion Scattering for Texture Classification
• Sparse arrays of signatures for online character recognition
• Persistence paths and signature features in topological data analysis
• The Signature Kernel is the solution of a Goursat PDE
• Time-warping invariants of multidimensional time series
• Signature Methods in Machine Learning
• Tropical time series, iterated-sums signatures and quasisymmetric functions
• Non-commutative Two Dimensional Modular Symbol
• Non-commutative Hilbert modular symbols
• 2-d signature of images and texture classification
• A Topological Approach to Mapping Space Signatures
• Two-parameter sums signatures and corresponding quasisymmetric functions
• Random Surfaces and Higher Algebra