What is Refractive Calibration?

The Problem

Standard multi-camera calibration assumes cameras and targets occupy the same optical medium. For underwater scenarios, this assumption breaks down: cameras are in air, viewing targets underwater through a flat water surface. Light rays from underwater targets refract at the air-water interface following Snell’s law. Standard calibration methods ignore refraction, leading to systematic errors in 3D reconstruction.

What AquaCal Does

AquaCal jointly optimizes:

• Camera extrinsics (position and orientation)

• Water surface position

• Calibration board poses

using refractive ray tracing to accurately model light paths through the interface. The calibration process uses standard ChArUco board observations but accounts for refraction during optimization.

Why It Matters

Without refractive modeling, underwater 3D reconstructions exhibit:

• Biased depth estimates (targets appear shallower than actual position)

• Systematic position errors increasing with distance from camera

• Breaking of geometric constraints (parallel lines appear non-parallel)

AquaCal provides accurate calibration for research applications requiring precise underwater measurements: behavioral tracking, environmental monitoring, volumetric capture.

Learn More

For detailed explanations of theory and implementation:

• Refractive Geometry — ray tracing through the water interface

• Coordinate Conventions — world frame, camera frame, transforms

• Optimizer Pipeline — bundle adjustment structure, parameters, and camera models

For practical usage:

• CLI Reference — command-line tools and options

• Troubleshooting — common issues and solutions

• User Guide — complete theory and practical guides

• API Reference — function and class documentation

• Tutorials — interactive notebook examples