Robust and Accurate Vision-Based Pose Estimation Algorithm Based on Four Coplanar Feature Points.

Area of Science:

• Computer Vision
• Robotics
• Machine Learning

Background:

• Vision-based pose estimation is crucial for machine vision applications.
• Existing analytical and iterative methods suffer from noise susceptibility and sensitivity to initial estimates, respectively.
• Image processing errors increase with measurement range, reducing pose estimation accuracy.

Purpose of the Study:

• To propose a novel pose estimation method that overcomes the limitations of existing techniques.
• To improve the accuracy and stability of object pose estimation, especially over larger measurement ranges.
• To reduce pose estimation errors through a new coordinate system approach.

Main Methods:

• A novel pose estimation method utilizing four coplanar points is introduced.
• Feature point coordinates are initially determined using linear constraints, followed by iterative optimization.
• A coordinate system for object motion is established to mitigate errors.

Main Results:

• The proposed method demonstrates improved efficiency and stability compared to two existing methods in experimental evaluations.
• The integration of linear and iterative optimization enhances the robustness of pose estimation.
• The established coordinate system effectively decreases pose estimation errors with increasing measurement range.

Conclusions:

• The novel four-coplanar-point method offers a more accurate and stable solution for vision-based pose estimation.
• This approach effectively addresses the limitations of traditional analytical and iterative methods.
• The proposed technique shows significant potential for enhancing machine vision systems.