Monocular Visual Pose Estimation Method Based on Spherical Cooperative Target

Area of Science:

• Computer Vision
• Robotics
• Metrology

Background:

• Conventional planar targets for visual measurement suffer from viewpoint changes and distortion.
• Spherical targets offer omnidirectional observability but PnP-based pose estimation can be error-prone with limited views or sparse features.

Purpose of the Study:

• To develop an integrated framework for high-precision spherical target construction and robust pose estimation.
• To overcome limitations of existing methods in close-range monocular visual measurement and cooperative target pose estimation.

Main Methods:

• Designed a composite marker layout with adaptively scaled latitude-longitude topology.
• Developed a center-to-pole point-cloud stitching strategy and fused multi-view observations for accurate target coordinate system reconstruction.
• Proposed a joint optimization method combining reprojection error with contour center consistency and introduced an SQPnP-based reinitialization mechanism.

Main Results:

• The proposed method demonstrated higher accuracy and robustness compared to EPnP, EPnP+LM, LM, and SQPnP in simulations.
• Achieved superior performance under continuous pose changes, sparse feature points, and varying noise levels.
• Real-image experiments confirmed the practical feasibility of the integrated framework.

Conclusions:

• The integrated framework provides a significant advancement in spherical target-based visual measurement and pose estimation.
• The method offers enhanced accuracy, robustness, and practical applicability for cooperative target pose estimation tasks.