Robust rigid registration via maximum correntropy criterion: a novel approach

Area of Science:

• Computer Vision
• Robotics
• Pattern Recognition

Background:

• Existing point set registration methods struggle with non-Gaussian noise, outliers, and incomplete data in complex engineering applications.
• Robustness is crucial for reliable performance in computer vision, pattern recognition, and intelligent robotics.

Purpose of the Study:

• To propose a novel robust rigid point set registration method.
• To improve registration accuracy and reliability in challenging real-world scenarios.

Main Methods:

• A new iterative optimization framework is developed based on the maximum correntropy criterion (MCC).
• The method integrates a point-to-plane distance metric with correntropy measurement.
• Nearest neighbor search establishes point correspondences, and rigid transformation parameters are optimized under the MCC criterion.
• An adaptive kernel width updating strategy is employed to balance global convergence and local alignment.

Main Results:

• Experimental results demonstrate the effectiveness of the proposed MCC-based method on both synthetic and real-world point sets.
• The method shows superior robustness against noise and outliers compared to existing approaches.
• The adaptive kernel width strategy effectively balances convergence speed and alignment precision.

Conclusions:

• The proposed maximum correntropy criterion (MCC)-based rigid point set registration method offers enhanced robustness and accuracy.
• This approach is well-suited for complex engineering scenarios involving noisy and incomplete point cloud data.
• The open-source code and data facilitate further research and application in computer vision and robotics.