Struct NdtConfig
pub struct NdtConfig {
pub max_iterations: usize,
pub resolution: f32,
pub epsilon: f64,
pub min_points_per_voxel: usize,
pub transformation_epsilon: f64,
pub fitness_epsilon: f64,
pub min_correspondences: usize,
pub initial_guess: Isometry3<f32>,
}Expand description
Configuration for NDT (Normal Distributions Transform) registration.
The target cloud is discretized into a voxel grid; each cell with enough points becomes a Gaussian (mean + covariance). The source is aligned by minimizing the Mahalanobis distance of each transformed point to its target cell distribution via Gauss-Newton (point-to-distribution NDT).
Fields§
§max_iterations: usizeMaximum number of optimization iterations.
resolution: f32Voxel size used to build the target distributions.
epsilon: f64Planar regularization: smallest eigenvalue floor as a fraction of the largest.
min_points_per_voxel: usizeMinimum points required for a target voxel to form a distribution.
transformation_epsilon: f64Stop when the transform update is smaller than this threshold.
fitness_epsilon: f64Stop when the fitness is smaller than this threshold.
min_correspondences: usizeMinimum number of matched points required per iteration.
initial_guess: Isometry3<f32>Initial transform guess mapping source into target frame.
Implementations§
§impl NdtConfig
impl NdtConfig
pub fn with_resolution(resolution: f32) -> NdtConfig
pub fn with_resolution(resolution: f32) -> NdtConfig
Creates a config with the given voxel resolution.