Detection and Classification of Artifact Distortions in Optical Motion Capture Sequences

Area of Science:

• Biomechanics and Motion Analysis
• Computer Vision and Image Processing

Background:

• Optical motion capture (OMC) systems are widely used but susceptible to marker recognition errors like occlusion and mislabeling.
• Post-capture software correction of these errors is imperfect, leading to residual artifactual distortions in the motion data.

Purpose of the Study:

• To examine four prevalent types of artifacts in optical motion capture data.
• To propose and evaluate a novel algorithmic method for the detection and classification of these motion distortions.

Main Methods:

• The proposed algorithm integrates derivative analysis, low-pass filtering, mathematical morphology, and a loose predictor.
• Testing involved simulations with synthetically distorted motion sequences.
• Performance was compared against human operators using real-world motion capture data.

Main Results:

• The developed algorithm demonstrated effectiveness in detecting and classifying motion artifact distortions.
• Comparisons indicated competitive or superior performance relative to human operators in specific scenarios.
• An analysis of applicability for artifact removal was also conducted.

Conclusions:

• The proposed derivative analysis-based algorithm offers a robust solution for identifying and categorizing artifacts in optical motion capture.
• This method has the potential to enhance the quality and reliability of motion data, aiding in subsequent distortion removal.