Crate vision_calibration 

High-level entry crate for the calibration-rs camera calibration library.

This crate provides a unified API for camera calibration workflows:

• Single-camera intrinsics calibration (Zhang’s method with distortion)
• Single-camera intrinsics calibration with Scheimpflug tilt
• Single-camera hand-eye calibration (camera on robot arm)
• Multi-camera rig extrinsics calibration
• Multi-camera rig + hand-eye calibration

Quick Start

use vision_calibration::prelude::*;
use vision_calibration::planar_intrinsics::{step_init, step_optimize};

// Create calibration session
let mut session = CalibrationSession::<PlanarIntrinsicsProblem>::new();
session.set_input(dataset)?;

// Option 1: Step-by-step (recommended for inspection)
step_init(&mut session, None)?;
step_optimize(&mut session, None)?;

// Option 2: Pipeline function (convenience)
// run_planar_intrinsics(&mut session)?;

// Export results
let result = session.export()?;

Module Organization

High-Level Calibration Workflows

• session - Session framework (CalibrationSession, ProblemType)
• planar_intrinsics - Single-camera intrinsics (Zhang’s method)
• single_cam_handeye - Single camera + hand-eye calibration
• rig_extrinsics - Multi-camera rig extrinsics
• rig_handeye - Multi-camera rig + hand-eye
• laserline_device - Single camera + laser plane device
• scheimpflug_intrinsics - Single-camera planar intrinsics with Scheimpflug tilt

Foundation Crates (Advanced Users)

• core - Math types, camera models, RANSAC primitives
• linear - Closed-form initialization algorithms
• optim - Non-linear least-squares optimization
• synthetic - Synthetic data generation for testing

Session API

All calibration workflows use the session::CalibrationSession state container. Each problem type has its own set of step functions — see the table below and the per-module documentation for details.

The common pattern is:

1. Create a session for the problem type
2. Set input data with set_input
3. Optionally configure with update_config
4. Run the problem-specific step functions (or a convenience run_calibration)
5. Export results with export

Available Problem Types

Problem Type	Input	Steps
PlanarIntrinsicsProblem	PlanarDataset	step_init → step_optimize
SingleCamHandeyeProblem	SingleCamHandeyeInput	step_intrinsics_init → step_intrinsics_optimize → step_handeye_init → step_handeye_optimize
RigExtrinsicsProblem	RigExtrinsicsInput	step_intrinsics_init_all → step_intrinsics_optimize_all → step_rig_init → step_rig_optimize
RigHandeyeProblem	RigHandeyeInput	step_intrinsics_init_all → step_intrinsics_optimize_all → step_rig_init → step_rig_optimize → step_handeye_init → step_handeye_optimize
LaserlineDeviceProblem	LaserlineDeviceInput	step_init → step_optimize
ScheimpflugIntrinsicsProblem	PlanarDataset	step_init → step_optimize

Modules

common

Step-option types shared across problem modules.

core

Core math types, camera models, and RANSAC primitives.

laserline_device

Single laserline device calibration (camera + laser plane).

linear

Closed-form initialization algorithms.

optim

Non-linear optimization vocabulary.

planar_intrinsics

Planar intrinsics calibration (Zhang’s method with Brown-Conrady distortion).

prelude

Minimal re-exports for planar “hello world” calibration.

rig_extrinsics

Multi-camera rig extrinsics calibration.

rig_handeye

Multi-camera rig hand-eye calibration.

rig_laserline_device

Rig-level laserline calibration.

scheimpflug_intrinsics

Single-camera planar intrinsics with Scheimpflug/tilted sensor refinement.

session

Session framework for structured calibration workflows.

single_cam_handeye

Single-camera hand-eye calibration (intrinsics + hand-eye transform).

synthetic

Deterministic synthetic data generation for testing.

Enums

Error

Typed error returned by all public calibration step functions. Errors returned by vision-calibration-pipeline public APIs.

Functions

compute_laserline_feature_residuals

Per-feature residual helpers from vision-calibration-optim re-exported for convenience: handeye_observer_se3_target (used by hand-eye exports) and compute_*_feature_residuals (used by laser exports). Compute per-pixel laser residual records for every laser observation in the dataset, indexed pose-major.

compute_rig_laserline_feature_residuals

Per-feature residual helpers from vision-calibration-optim re-exported for convenience: handeye_observer_se3_target (used by hand-eye exports) and compute_*_feature_residuals (used by laser exports). Compute per-pixel laser residual records for every laser observation in a rig dataset, indexed pose-major then by camera then by pixel.

handeye_observer_se3_target

Per-feature residual helpers from vision-calibration-optim re-exported for convenience: handeye_observer_se3_target (used by hand-eye exports) and compute_*_feature_residuals (used by laser exports). Derive per-view “observer-to-target” poses from a hand-eye chain.

pixel_to_gripper_point

Map a laser pixel in a specific camera to a 3D point in the robot gripper frame. Map a laser pixel in a specific camera to a 3D point in the robot gripper frame.