Three-dimensional mandibular motion trajectory-tracking system based on BP neural network.

Area of Science:

• Biomedical Engineering
• Computer Vision
• Neural Networks

Background:

• Accurate real-time mandibular movement recording is crucial for diagnosing oral diseases and digital restoration design.
• Existing 3D motion capture systems face challenges with calibration complexity and measurement accuracy.
• Involuntary physiological motions (heartbeat, respiration) can degrade tracking precision.

Purpose of the Study:

• To develop a prototype 3D optical motion capture system for accurate and real-time mandibular movement recording.
• To implement a 3D compensation method to eliminate involuntary physiological motion artifacts.
• To enhance camera calibration and 3D measurement accuracy using binocular stereo vision and Back-propagation Neural Networks.

Main Methods:

• Developed a binocular stereo vision system integrated with a Back-propagation (BP) Neural Network.
• Implemented a specialized 3D compensation technique to filter out vibrations from heartbeat and respiration.
• Utilized a binocular visual 3D measurement method based on projection lines and a BP neural network-based calibration approach.

Main Results:

• The system demonstrated high accuracy, with a root-mean-square error of 0.0773 mm, validated through electric platform and clinical trials.
• The proposed BP neural network calibration method simplified the process and improved 3D measurement accuracy.
• Comparisons showed the developed system offers superior reliability and accuracy compared to existing state-of-the-art methods.

Conclusions:

• The prototype 3D optical motion capture system provides accurate and reliable real-time mandibular movement tracking.
• The system's advanced compensation and calibration methods effectively address key challenges in 3D motion capture.
• This technology holds significant potential for improving the diagnosis of oral diseases and advancing digital dental restoration design.