Struct XJunctionValidator 

pub struct XJunctionValidator<F: Float = f32> {
    pub min_spacing: F,
    pub max_spacing: F,
    pub tolerance_rad: F,
}

Validator for square grids of X-junction corners with known orientation.

Designed for ChESS-like features where each corner has an orientation angle (mod π) and adjacent corners on a chessboard pattern have orthogonal orientations. The edge between two neighbors is at ~45° to both orientations.

PointData

F — the corner’s orientation angle in radians (mod π, undirected axis).

Example

use projective_grid::{GridGraph, GridGraphParams, NeighborCandidate};
use projective_grid::square::validators::XJunctionValidator;
use nalgebra::Point2;
use std::f32::consts::FRAC_PI_4;

let positions = vec![
    Point2::new(0.0f32, 0.0), Point2::new(10.0, 0.0),
    Point2::new(0.0, 10.0),   Point2::new(10.0, 10.0),
];
// Alternating orientations (π/4 and 3π/4) like a chessboard
let orientations = vec![FRAC_PI_4, 3.0 * FRAC_PI_4, 3.0 * FRAC_PI_4, FRAC_PI_4];

let validator = XJunctionValidator {
    min_spacing: 5.0,
    max_spacing: 15.0,
    tolerance_rad: 15.0f32.to_radians(),
};
let graph = GridGraph::build(
    &positions, &orientations, &validator, &GridGraphParams::default(),
);

Fields

min_spacing: F

Minimum acceptable neighbor distance (pixels).

max_spacing: F

Maximum acceptable neighbor distance (pixels).

tolerance_rad: F

Angular tolerance (radians) for orthogonality and 45° edge alignment checks.

Trait Implementations

impl<F: Float> NeighborValidator<F> for XJunctionValidator<F>

type PointData = F

Per-point data beyond position (e.g., orientation, cluster label). Use () if no extra data is needed.

fn validate( &self, _source_index: usize, source_data: &F, candidate: &NeighborCandidate<F>, candidate_data: &F, ) -> Option<(NeighborDirection, F)>

Validate whether candidate is a valid grid neighbor of the point at source_index. Returns (direction, score) where lower score is better, or None to reject.