Respiratory Monitoring Based on Video Processing

Abstract

A modified method to extract and track a person’s subtle respiratory movement during sleep based on video sequence is proposed. To detect this subtle movement, we need accurately register the video sequence. After registration, Lucas-Kanade optical flow was adopted to track feature points. By analyzing these feature trajectories, feature points that represent the respiratory movement during breathing are extracted. The experiment results show that this method could successfully track the respiratory movement of a person with lying posture.