Extrinsic Parameter Calibration for Line Scanning Cameras on Ground Vehicles with Navigation Systems Using a...

Area of Science:

• Robotics and Computer Vision
• Geospatial Data Acquisition

Background:

• Line scanning cameras are increasingly used in mobile and robotic platforms.
• Accurate 6D pose estimation is crucial for direct georeferencing of camera data.
• Mobile platforms often have navigation systems providing vehicle pose, but camera pose relative to this system may be unknown.

Purpose of the Study:

• To develop a novel method for estimating the 6D pose of a rigidly mounted line scanning camera relative to a mobile platform's navigation system.
• To provide accurate georeferencing capabilities for line scanning camera data.

Main Methods:

• A calibration pattern with identifiable points is imaged and manually labeled.
• Points are triangulated using data from both the line scanning camera and the navigation system.
• A likelihood function based on point reprojection is maximized to estimate the 6D camera pose.
• A Markov Chain Monte Carlo (MCMC) algorithm is employed to quantify the uncertainty of the estimated pose offset.

Main Results:

• The proposed method successfully estimated the camera's 6D pose on two different platforms.
• Pose estimation accuracy achieved was within 0.06 m/1.05 degrees and 0.18 m/2.39 degrees.
• The study also explored methods for human-readable visualization and interpretation of the 6D pose results.

Conclusions:

• The novel method provides accurate 6D pose estimation for line scanning cameras on mobile platforms.
• This technique significantly improves the direct georeferencing of sensor data.
• The approach is robust and applicable to various robotic and ground-based systems.