Camera-LiDAR Wide Range Calibration in Traffic Surveillance Systems

Area of Science:

• Computer Vision
• Robotics
• Sensor Fusion

Background:

• Accurate camera-LiDAR calibration is essential for traffic surveillance systems.
• Large surveillance areas increase calibration complexity due to expanded sensor distances and search spaces.
• Existing methods may struggle with the complexity of wide-area calibration.

Purpose of the Study:

• To propose a novel target-less camera-LiDAR calibration method for traffic surveillance.
• To address the challenges of large search spaces in wide-area calibration.
• To enhance calibration accuracy and robustness in complex environments.

Main Methods:

• Leveraging dynamic objects (moving vehicles) for calibration.
• Employing a genetic algorithm-based optimization technique to constrain the search range.
• Developing a target-less approach to simplify the calibration process.

Main Results:

• Achieved high calibration accuracy in experimental results.
• Demonstrated suitability for wide-area traffic surveillance applications.
• Successfully constrained the calibration search range using dynamic objects.

Conclusions:

• The proposed target-less method effectively calibrates camera-LiDAR systems for traffic surveillance.
• Genetic algorithm optimization mitigates risks of local optima convergence.
• This approach offers a promising solution for enhancing sensor fusion in complex surveillance settings.