Abstract:The aim of this paper is to provide a coherent framework for transforming boundary pairs of digital images from one resolution to another without knowledge of the full images. It is intended to facilitate the simultaneous usage of multiresolution processing and boundary reduction, primarily for algorithms in computational dynamics and computational control theory.
What problem does this paper attempt to address?
The problem that this paper attempts to solve is to provide a coherent framework for the conversion between different resolutions of boundary pairs of digital images without the need to know complete image information. Specifically, the author aims to promote the simultaneous use of multi - resolution processing and boundary reduction techniques, especially in algorithms of computational dynamics and computational control theory.
### Main problems and objectives of the paper
1. **Problem background**:
- Digital images are important data structures in many fields, such as computational dynamics, computational control theory, and rigorous computation.
- In these fields, algorithms usually start with rough area coverage and improve accuracy by gradually refining and eliminating unnecessary parts.
- The boundary - tracking algorithm reduces the complexity of reachable set calculations by only storing and operating on boundary pairs (i.e., the boundaries of digital images and their adjacent layers).
2. **Main objectives**:
- Provide a coherent framework for converting boundary pairs of digital images from one resolution to another without complete image information.
- Promote the simultaneous use of multi - resolution processing and boundary reduction techniques, especially in computational dynamics and computational control theory.
### Specific contributions
The main contributions of the paper include:
- **Explore the geometric structure of digital images**: The inner and outer layer structures of digital images are studied, especially their relationship with the distance transformation.
- **Characterize the boundary - pair space of all digital images**: Define and describe the space of all possible boundary pairs.
- **Identify restriction operators and interpolation operators**: Propose a restriction operator \(R\) and an interpolation operator \(I\) that can project digital images on a fine grid onto a coarse grid and satisfy a series of desirable properties.
- **Lift operators to the boundary - pair space**: Lift the operators \(R\) and \(I\) to mappings \(\partial R\) and \(\partial I\) between boundary - pair spaces, and provide implementation methods that do not depend on complete image information.
### Formula representation
To ensure the correctness and readability of the formulas, the following are some key formulas involved in the paper:
- **Hausdorff semi - distance and full distance**:
\[
d(M, M') := \sup_{x \in M} \inf_{x' \in M'} \|x - x'\|_\infty
\]
\[
d_H(M, M') := \max \{d(M, M'), d(M', M)\}
\]
- **Definition of path**:
\[
p = (\xi_0, \xi_1,\ldots, \xi_k) \in \Delta_\rho^{k + 1}
\]
where \(k\in\mathbb{N}\), satisfying \(\xi_0 = x\), \(\xi_k = z\), and
\[
\|\xi_\ell-\xi_{\ell - 1}\|_\infty\leq\rho\quad\forall\ell\in(0, k]
\]
- **Definition of boundary pair**:
\[
\partial_0^\rho M := \{x\in M : \exists z\in M^c\cap\Delta_\rho\text{ with }\|x - z\|_\infty=\rho\}
\]
\[
\partial_k^\rho M := \{z\in M^c\cap\Delta_\rho : d(z, \partial_0^\rho M) = k\rho\},\quad k\in\mathbb{N}
\]
Through these formulas and definitions, the paper provides a solid theoretical basis for the conversion of boundary pairs of digital images between different resolutions.