Struct Obs

pub struct Obs<'a> {
    pub intrinsics: &'a Intrinsics,
    pub template: &'a DMatrix<u8>,
    pub image: &'a DMatrix<u8>,
    pub coordinates: &'a Vec<(usize, usize)>,
    pub _z_candidates: &'a Vec<Float>,
    pub jacobians: &'a Vec<Vec6>,
    pub hessians: &'a Vec<Mat6>,
}

Precomputed data available for the optimizer iterations:

Fields

intrinsics: &'a Intrinsics

Intrinsic parameters of the camera.

template: &'a DMatrix<u8>

Reference (“keyframe”) image.

image: &'a DMatrix<u8>

Current image to track.

coordinates: &'a Vec<(usize, usize)>

Coordinates of the points used for the tracking.

_z_candidates: &'a Vec<Float>

Inverse depth of the points used for the tracking.

jacobians: &'a Vec<Vec6>

Jacobians precomputed for the points used for the tracking.

hessians: &'a Vec<Mat6>

Hessian matrices precomputed for the points used for the tracking.

Trait Implementations

impl<'a> State<Obs<'a>, Result<EvalData, f32>, Isometry<f32, U3, Unit<Quaternion<f32>>>, String> for LMOptimizerState

impl<'a> optimizer::State<Obs<'a>, EvalState, Iso3, String> for LMOptimizerState.

fn init(obs: &Obs<'_>, model: Iso3) -> Self

Initialize the optimizer state.

fn step(&self) -> Result<Iso3, String>

Compute the step using Levenberg-Marquardt. Apply the step in an inverse compositional approach to compute the next motion estimation. May return an error at the Cholesky decomposition of the hessian.

fn eval(&self, obs: &Obs<'_>, model: Iso3) -> EvalState

Compute residuals and energy of the new model. Then, evaluate the new hessian and gradient if the energy has decreased.

fn stop_criterion( self, nb_iter: usize, eval_state: EvalState, ) -> (Self, Continue)

Stop after too many iterations, or if the energy variation is too low.

Also update the Levenberg-Marquardt coefficient depending on if the energy increased or decreased.

fn iterative_solve( obs: &Observations, initial_model: Model, ) -> Result<(Self, usize), Error>

Iteratively solve your optimization problem, with the provided functions by the trait implementation. Return the final state and the number of iterations. May return an error if a step computation failed.