Rotation-Invariant Edge Index

• Rotation-Invariant Edge Index is a family of techniques that maintain edge or graph descriptors invariant to rotation across different domains.
• In multimodal remote sensing, it normalizes orientation channels from Log-Gabor responses to enhance keypoint matching under rotational changes.
• In point-cloud and graph analyses, it encodes invariant connectivity and edge attributes, fostering reliable shape recognition and robust matching.

Searching arXiv for the cited works to ground the article in current records. Searching arXiv for multimodal remote sensing and point-cloud papers. In contemporary arXiv literature, the expression rotation-invariant edge index denotes a family of constructions rather than a single formal object. In multimodal remote sensing, it refers to a normalized orientation-channel index derived from Log-Gabor edge responses and used for keypoint description; in point-cloud and geometric deep learning, it usually denotes a rotation-invariant graph connectivity structure together with invariant edge attributes; in shape analysis, it appears as an edge- or boundary-based transform whose coefficients or magnitudes are invariant to in-plane rotation; and in chemical graph theory and discrete geometry, it denotes edge-defined invariants whose values are unchanged by planar embedding rotation or rigid body motion (Zhu et al., 2022, Yu et al., 2023, Črepnjak et al., 2017).

1. Terminological scope and core definitions

A first distinction is between edge index as a local orientation label and edge index as graph connectivity. In the remote-sensing formulation of R2FD2, the Edge Index at pixel $(x,y)$ is the orientation-channel index where the edge-like response is maximal,

$$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$

with $A_o(x,y)=\sum_s M_{s,o}(x,y)$ obtained from multi-scale Log-Gabor magnitudes. Rotation invariance is achieved by subtracting a patchwise dominant orientation index and rotating the local patch to a canonical orientation (Zhu et al., 2022).

In point-cloud learning, by contrast, an edge index is a set of directed pairs $(i,j)$ indicating graph connectivity. In common frameworks such as PyTorch Geometric, edge_index is a $2\times E$ integer tensor of source and target node indices. Rotation invariance then does not come from relabeling orientation channels, but from constructing neighborhoods through Euclidean $k$-NN or radius graphs and attaching pose-invariant edge attributes such as Point Pair Features (PPF) (Yu et al., 2023).

A third usage appears in distance-based graph invariants. For benzenoid systems, the edge-Wiener index, edge-Szeged index, and PI index depend only on graph-theoretic distances and partition counts. Because they are defined on graphs rather than embeddings, a rotation of the planar drawing is only a relabeling of vertices and edges, so the resulting indices are rotation-invariant with respect to planar embeddings (Črepnjak et al., 2017).

This terminological spread is important because many apparent disagreements in the literature arise from the fact that “edge index” may denote a discrete orientation label, a graph adjacency tensor, or an edge-defined invariant scalar. A plausible implication is that comparisons across domains are meaningful only at the level of invariance mechanisms, not at the level of a single shared data structure.

2. Rotation-invariant edge indexing in multimodal remote sensing

The most literal use of the term appears in R2FD2, a two-stage pipeline for multimodal remote sensing image matching. Stage 1 is the MALG detector, which computes Log-Gabor responses at multiple scales and orientations and forms a multi-channel auto-correlation saliency measure. Stage 2 is the RMLG descriptor, which first assigns a dominant orientation through a Rotation-invariant Maximum Index Map (RMIM) and then encodes the resulting normalized map with DAISY’s spatial configuration (Zhu et al., 2022).

The underlying filter bank is defined in polar frequency coordinates by

$$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$

with multi-scale, multi-orientation responses

$$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$

After aggregation across scales,

$A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$

the Maximum Index Map is

$$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$

This MIM is the paper’s Edge Index.

Patchwise normalization proceeds by taking the mode of MIM values in a local neighborhood around an interest point,

$$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$0

mapping it to a dominant orientation

$$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$1

re-indexing the orientation channels by circular shift,

$$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$2

and rotating the patch to obtain RMIM. In the equivalent “normalized Edge Index” form, the paper defines

$$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$3

followed by patch rotation by $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$4; this compensates both the shift in dominant orientation channel and the spatial rotation induced by a global image rotation (Zhu et al., 2022).

The descriptor stage uses DAISY’s circular radial grid with three rings, eight samples per ring, and one center bin. With $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$5 orientation bins, the descriptor dimensionality is $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$6. Histogram bins are formed over RMIM indices,

$$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$7

optionally amplitude-weighted, concatenated, and typically $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$8-normalized. Because dominant orientation is assigned from the mode of a discrete index map, the procedure avoids the interpolation-heavy orientation search associated with SIFT-like designs (Zhu et al., 2022).

The same paper explicitly links this construction to multimodal robustness. Radiation variations across optical, infrared, LiDAR intensity images, and SAR are described as severe, while Log-Gabor and phase-derived features are described as more stable than raw gradients because they isolate structural information and are less affected by absolute intensities. Experimentally, MALG repeatability is reported as 57.38% vs 51.50% on optical-infrared, 40.32% vs 35.42% on optical-LiDAR, and 34.14% vs 27.84% on optical-SAR relative to FASTMPC. In rotation sweeps from 0° to 360° in 10° steps, the number of correct matches is reported as never below 100 and often above 300, with 100% success rate; across datasets, R2FD2 reports RMSE < 2 pixels and runtime ≈ 11.27 s per dataset batch, faster than OS-SIFT, RIFT, HOSS, MS-HLMO, and RI-ALGH (Zhu et al., 2022).

3. Point-cloud graphs, local frames, and invariant edge attributes

In point-cloud matching and analysis, a rotation-invariant edge index is usually a graph whose connectivity is preserved under rigid transforms and whose edge features are pose-agnostic. RoITr states this explicitly: for a point set $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$9, the edge index is a set of directed pairs $A_o(x,y)=\sum_s M_{s,o}(x,y)$0 indicating that point $A_o(x,y)=\sum_s M_{s,o}(x,y)$1 is a neighbor of point $A_o(x,y)=\sum_s M_{s,o}(x,y)$2, and Euclidean $A_o(x,y)=\sum_s M_{s,o}(x,y)$3-NN neighborhoods are invariant under rotation and translation because distances are preserved. RoITr attaches to each edge the Point Pair Feature

$A_o(x,y)=\sum_s M_{s,o}(x,y)$4

where $A_o(x,y)=\sum_s M_{s,o}(x,y)$5 and $A_o(x,y)=\sum_s M_{s,o}(x,y)$6. Each component is unchanged by rigid rotation, so connectivity and edge attributes are both rotation-invariant up to $A_o(x,y)=\sum_s M_{s,o}(x,y)$7-NN tie-breaking (Yu et al., 2023).

RoITr embeds these PPF coordinates inside attention. In the local PPF Attention Mechanism, linearly embedded geometry terms $A_o(x,y)=\sum_s M_{s,o}(x,y)$8 and $A_o(x,y)=\sum_s M_{s,o}(x,y)$9 are combined with projected features $(i,j)$0, $(i,j)$1, and $(i,j)$2 through

$(i,j)$3

The architecture stacks Attentional Abstraction Layers and PPF Attention Layers, and then adds a global transformer with a Geometry-Aware Self-Attention Module and a Position-Aware Cross-Attention Module. The stated motivation is that finite rotation augmentation cannot span continuous $(i,j)$4, whereas PPF-based coordinates provide intrinsic invariance across the continuous rotation manifold. On Rotated 3DLoMatch, RoITr reports IR 53.2% vs GeoTrans 40.0%, RR 77.2% vs GeoTrans 71.8%, and FMR 89.4% vs GeoTrans 85.8% (Yu et al., 2023).

A different strategy appears in the PCA-based framework for rotation-invariant point-cloud analysis. There, centered coordinates $(i,j)$5 are mapped into canonical coordinates $(i,j)$6 after eigendecomposition of the covariance $(i,j)$7. Because sorted eigenvectors still admit sign ambiguity, the framework retains four valid frames

$(i,j)$8

and resolves the ambiguity through permutation-invariant fusion across the four branches. Neighborhoods built from Euclidean distances remain rotation-invariant either in the original or canonical coordinates (Luo et al., 2024).

AECNN obtains the same goal through Local Reference Frames rather than global PCA. For a reference point $(i,j)$9, it computes a neighbor barycenter

$2\times E$0

sets the local $2\times E$1-axis to

$2\times E$2

projects $2\times E$3 onto the plane orthogonal to $2\times E$4, and defines an orthonormal frame $2\times E$5. Neighbors are represented by

$2\times E$6

and aligned edge convolution uses

$2\times E$7

followed by

$2\times E$8

On ModelNet40, AECNN reports 91.0 under Y/Y, Y/AR, and AR/AR; on ShapeNet Part it reports 80.2 mIoU under the same rotation-perturbed protocols (Zhang et al., 2021).

SpinConv uses a per-edge local frame. For edge $2\times E$9 it aligns

$k$0

to the global $k$1-axis with a rotation matrix $k$2, maps neighbor directions into that frame, writes them in spherical coordinates $k$3, performs a circular spin convolution along the azimuthal dimension, and then average-pools over all circular shifts. The key invariant is therefore not the raw aligned grid, but the pooled output

$k$4

which eliminates the residual roll angle around the edge direction. The energy-centric model then predicts

$k$5

so scalar energy is rotation-invariant by construction (Shuaibi et al., 2021).

4. Edge-based transforms for images and planar shapes

A separate line of work constructs rotation-invariant indices from boundary or edge geometry itself. The 2026 edge-based Zernike formulation replaces pixel-center quadrature by exact line integrals over image boundaries. Zernike moments are

$k$6

and under rotation by $k$7 satisfy

$k$8

so $k$9 is rotation-invariant. The edge-based derivation applies Green’s theorem, introduces the radial primitive

$$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$0

and reduces each directed boundary edge to a one-dimensional integral. After parameterization by $$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$1, the transformed integrand $$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$2 is polynomial in $$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$3 of degree at most $$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$4, so Clenshaw–Curtis quadrature with $$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$5 evaluates it exactly. The resulting index is the collection of $$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$6, optionally normalized by $$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$7 (Koehl et al., 13 Jul 2026).

PSepT uses a different edge-based polar transform. It defines a tensor-product basis on a discrete polar grid with a DCT-II radial basis

$$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$8

and discrete Fourier angular harmonics

$$G(\rho,\theta)=\exp\!\left(-\frac{\left[\ln(\rho/\rho_0)\right]^2}{2\left[\ln(\sigma_\rho)\right]^2}\right)\cdot \exp\!\left(-\frac{(\theta-\theta_0)^2}{2\sigma_\theta^2}\right),$$9

Coefficients

$$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$0

obey the rotation-covariance relation

$$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$1

Therefore $$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$2 and

$$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$3

are rotation-invariant descriptors. The paper states complexity $$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$4, memory $$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$5, and condition number scaling $$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$6, contrasting these with polynomial moment families such as Zernike and pseudo-Zernike (Singh et al., 10 Oct 2025).

For standardized star-shaped planar objects, the vectorized shape representation paper encodes the boundary as a radial function $$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$7 on $$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$8 and defines the discrete rotation-invariant sketch

$$R_{s,o}(x,y)=(I*L_{s,o})(x,y), \qquad M_{s,o}(x,y)=|R_{s,o}(x,y)|.$$9

with $A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$0 in the central analysis. This sketch is invariant to rotation and constant offset, and also invariant under the reverse-of-complement (RoC) operation. The same paper states an FFT-based computation in $A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$1 and a deterministic $A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$2-approximation with dimension $A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$3 for standardized star-shaped objects (Shafieasl et al., 26 May 2026).

These three formulations share a common principle: boundary or edge geometry is converted into coefficients that are rotation-covariant before magnitude/pooling and rotation-invariant after magnitude/pooling. This suggests a unifying distinction between covariant intermediate representations and invariant final indices.

5. Mesh coordinates and graph-theoretic edge indices

In discrete surface geometry, the Nonlinear Rotation-Invariant Coordinates framework defines a rotation-invariant edge index directly on a triangular mesh. For every edge $A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$4,

$A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$5

and for every interior edge the signed dihedral angle is

$A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$6

where

$A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$7

The rotation-invariant edge index is therefore the map $A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$8 for interior edges, with boundary edges carrying only $A_o(x,y)=\sum_{s=1}^{S} M_{s,o}(x,y),$9. The framework states that matching $$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$0 and $$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$1 for all edges determines an immersion up to $$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$2 rigid motion (Sassen et al., 2019).

The same work characterizes valid edge-index vectors by local triangle inequalities and quaternion-valued integrability constraints around interior vertices. In aligned face frames, adjacent faces satisfy

$$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$3

and each interior vertex imposes

$$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$4

The image of the NRIC map is therefore described implicitly as

$$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$5

This construction is used for near-isometric deformation, time-discrete geodesics, and reconstruction from almost-integrable edge coordinates (Sassen et al., 2019).

In chemical graph theory, the expression “edge index” has an older and more abstract meaning. The edge-Wiener index is

$$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$6

where edge–edge distance is measured in the line graph. The edge-Szeged index is

$$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$7

and the edge PI index is

$$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$8

For benzenoid systems, these indices are rotation-invariant with respect to planar embeddings because they depend only on graph distances and counts, and the paper shows that they can be computed in time $$\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y).$$9, where $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$00 is the boundary cycle (Črepnjak et al., 2017).

The mesh and graph-theoretic cases illustrate a different notion of invariance from the image and point-cloud literature. Here the edge index is itself the invariant object, not merely an intermediate representation used to build a descriptor.

6. Recurring principles, limitations, and misconceptions

Across these literatures, four recurrent mechanisms produce rotation invariance. The first is canonical reindexing or canonical orientation assignment, as in RMIM, where orientation channels are circularly shifted so that the dominant channel becomes index $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$01 and the patch is rotated by the dominant orientation (Zhu et al., 2022). The second is construction from rigid-motion invariants, such as Euclidean distances, centered inner products, edge lengths, and PPF angles (Yu et al., 2023, Luo et al., 2024). The third is local-frame alignment followed by pooling over the residual degree of freedom, which is the central pattern in AECNN and SpinConv (Zhang et al., 2021, Shuaibi et al., 2021). The fourth is magnitude or power-spectrum extraction from rotation-covariant coefficients, used by Zernike moments, PSepT, and the vectorized radial sketch (Koehl et al., 13 Jul 2026, Singh et al., 10 Oct 2025, Shafieasl et al., 26 May 2026).

Several limitations recur just as consistently. R2FD2 explicitly reports no explicit scale invariance, orientation quantization with $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$02 and $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$03, weaker responses in low-texture regions, and failure modes under extreme multimodal disparities such as clouds or severe SAR speckle (Zhu et al., 2022). RoITr and related point-cloud methods depend on normal estimation; the supplied material notes that noisy or ambiguous normals, degenerate geometries, and sparse point clouds reduce PPF distinctiveness and can destabilize attention (Yu et al., 2023). The PCA framework identifies eigenvalue degeneracy and sign ambiguity as fundamental, not incidental, which is why it replaces single-frame canonicalization by a four-frame set with permutation-invariant fusion (Luo et al., 2024). The vectorized planar-shape representation introduces an intrinsic RoC ambiguity, which must be controlled by canonicalization or by storing both variants (Shafieasl et al., 26 May 2026). PSepT identifies centroid errors, partial occlusions, and edge fragmentation as practical sources of instability even when the transform itself is rotation-covariant (Singh et al., 10 Oct 2025). NRIC, finally, increases the number of variables and adds equality constraints $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$04, so its rotation invariance is accompanied by a nontrivial integrability problem (Sassen et al., 2019).

A common misconception is that rotation invariance is equivalent to training with rotated data. RoITr states the opposite motivation explicitly: finite rotation augmentation cannot span continuous $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$05, so invariance obtained only extrinsically remains unstable under rarely seen rotations (Yu et al., 2023). Another misconception is that a rotation-invariant edge index automatically handles all nuisance transformations. The supplied material repeatedly separates rotation from radiometric change, translation, scale, and topology: R2FD2 is designed for radiation and rotation robustness but not full scale invariance; point-cloud graphs built from Euclidean $$\mathrm{EI}(x,y)=\mathrm{MIM}(x,y)=\underset{o}{\arg\max}\;A_o(x,y),$$06-NN are rotation- and translation-invariant but not inherently robust to poor normals or non-rigid distortion; and shape signatures based on radial functions require centering and scale normalization before their rotation invariance becomes meaningful (Zhu et al., 2022, Yu et al., 2023, Shafieasl et al., 26 May 2026).

Taken together, these formulations show that the modern “rotation-invariant edge index” is best understood as a domain-dependent invariant encoding attached to edge structure. In some settings the index is a normalized orientation label, in others a rigid-motion-invariant graph relation, and in others an edge-derived invariant statistic or coordinate chart. What unifies them is not a single implementation, but the requirement that edge-centered structure remain stable under rotation while preserving enough information for matching, retrieval, optimization, or recognition.