ORCA: Oscillating-Resonant Component Analysis for Video-Based Respiration Monitoring

Current respiration monitoring systems mostly rely on multiple devices that are directly attached to the body. These devices are both expensive and elaborate, requiring medical professionals to install and maintain. Some previous studies explored the advantages of video-based respiration monitoring systems, but they failed to provide a complete solution for long-term monitoring, and other production-related problems such as environmental setup and noises. To overcome these difficulties, we introduce a lightweight yet effective algorithm to measure respiration information by extracting oscillating components from optical flow analysis. Extensive experiments showed remarkable impact of the proposed technique, notably it successfully detected the respiration signal in 5/6 videos with severe blinking problems, thus demonstrating the potential to be applied in large scale to the medical field.