Struct FThetaCamera 

pub struct FThetaCamera {
    pub cx: f64,
    pub cy: f64,
    pub distortion: DistortionModel,
}

NVIDIA f-theta fisheye camera with 6 intrinsic parameters.

Stores the principal point (cx, cy) and the four forward-polynomial coefficients k₁…k₄. The model is isotropic (no separate fx/fy).

Fields

cx: f64

Principal point x (u₀ in the paper), pixels.

cy: f64

Principal point y (v₀ in the paper), pixels.

distortion: DistortionModel

Forward-polynomial distortion — must be DistortionModel::FTheta.

Implementations

impl FThetaCamera

pub fn new( cx: f64, cy: f64, distortion: DistortionModel, ) -> Result<Self, CameraModelError>

Creates a new f-theta camera.

Errors

Returns CameraModelError::InvalidParams if distortion is not DistortionModel::FTheta, or if any parameter fails validation.

Example

use apex_camera_models::{CameraModel, DistortionModel, FThetaCamera};

let distortion = DistortionModel::FTheta { k1: 500.0, k2: 0.0, k3: 0.0, k4: 0.0 };
let camera = FThetaCamera::new(320.0, 240.0, distortion)?;
assert_eq!(camera.get_model_name(), "ftheta");

pub fn try_from_params(params: &[f64]) -> Result<Self, CameraModelError>

Build directly from the six scalar parameters [cx, cy, k1, k2, k3, k4].

pub fn distortion_params(&self) -> (f64, f64, f64, f64)

Extract (k1, k2, k3, k4) from the stored distortion.

Panics

Panics if self.distortion is not DistortionModel::FTheta — this cannot happen after successful construction.

pub fn linear_estimation( &self, points_3d: &Matrix3xX<f64>, points_2d: &Matrix2xX<f64>, ) -> Result<Self, CameraModelError>

Least-squares estimation of k₁..k₄ given 3D-2D correspondences, assuming cx and cy are already known.

Constructs a Vandermonde system [θ θ² θ³ θ⁴] · [k₁ k₂ k₃ k₄]ᵀ = r and solves it with SVD (nalgebra full-pivoting).

Returns an updated FThetaCamera with the estimated polynomial coefficients. The principal point is unchanged.

Arguments

• points_3d — 3×N matrix of 3D points in camera frame
• points_2d — 2×N matrix of observed pixel coordinates

Trait Implementations

impl CameraModel for FThetaCamera

const INTRINSIC_DIM: usize = 6

Number of intrinsic parameters (compile-time constant).

type IntrinsicJacobian = Matrix<f64, Const<2>, Const<6>, ArrayStorage<f64, 2, 6>>

Jacobian type for intrinsics: 2 × INTRINSIC_DIM.

type PointJacobian = Matrix<f64, Const<2>, Const<3>, ArrayStorage<f64, 2, 3>>

Jacobian type for 3D point: 2 × 3.

fn project( &self, p_cam: &Vector3<f64>, ) -> Result<Vector2<f64>, CameraModelError>

Projects a 3D point in camera coordinates to 2D image coordinates. See the cookbook introduction for the projection pipeline, and the per-model page for the formula.

fn unproject( &self, point_2d: &Vector2<f64>, ) -> Result<Vector3<f64>, CameraModelError>

Unprojects a 2D image point to a normalized 3D ray in camera frame. Some models use Newton-Raphson for undistortion. See the per-model cookbook page for the algorithm.

fn jacobian_point(&self, p_cam: &Vector3<f64>) -> Self::PointJacobian

∂(u,v)/∂(x,y,z) — 2×3 Jacobian of projection w.r.t. the 3D point. Used for structure optimisation, triangulation, and bundle adjustment. See the per-model cookbook page for the formula.

fn jacobian_intrinsics(&self, p_cam: &Vector3<f64>) -> Self::IntrinsicJacobian

∂(u,v)/∂(params) — 2×N Jacobian of projection w.r.t. intrinsic parameters, where N = INTRINSIC_DIM. The parameter order is model-specific; see the per-model cookbook page.

fn validate_params(&self) -> Result<(), CameraModelError>

Validates camera intrinsic and distortion parameters. The exact rules are model-specific; see the per-model cookbook page under “Validation Rules”.

fn get_pinhole_params(&self) -> PinholeParams

Returns the linear intrinsics (f_x, f_y, c_x, c_y).

fn get_distortion(&self) -> DistortionModel

Returns the distortion model and its parameters.

fn get_model_name(&self) -> &'static str

Returns the camera model identifier ("pinhole", "rad_tan", …).

fn jacobian_pose( &self, p_world: &Vector3<f64>, pose: &SE3, ) -> (Self::PointJacobian, SMatrix<f64, 3, 6>)

∂(u,v)/∂(δξ) — 2×6 Jacobian of projection w.r.t. the camera pose.

fn project_batch(&self, points_cam: &Matrix3xX<f64>) -> Matrix2xX<f64>

Projects N 3D points in one call. Invalid projections are replaced by the sentinel (1e6, 1e6). Models may override with a vectorised implementation.

impl Clone for FThetaCamera

fn clone(&self) -> FThetaCamera

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Copy for FThetaCamera

impl Debug for FThetaCamera

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl From<&FThetaCamera> for [f64; 6]

Parameter order: [cx, cy, k1, k2, k3, k4].

fn from(cam: &FThetaCamera) -> Self

Converts to this type from the input type.

impl From<&FThetaCamera> for DVector<f64>

fn from(cam: &FThetaCamera) -> Self

Converts to this type from the input type.

impl From<[f64; 6]> for FThetaCamera

fn from(p: [f64; 6]) -> Self

Converts to this type from the input type.

impl PartialEq for FThetaCamera

fn eq(&self, other: &FThetaCamera) -> bool

Equality operator ==.

fn ne(&self, other: &Rhs) -> bool

Inequality operator !=.

impl StructuralPartialEq for FThetaCamera

impl TryFrom<&[f64]> for FThetaCamera

type Error = CameraModelError

The type returned in the event of a conversion error.

fn try_from(params: &[f64]) -> Result<Self, Self::Error>

Performs the conversion.