부산대학교 도서관

Video based heart rate detection performs well for monitoring human in still or small motion case, such as patients. When targets are in intensive exercise, strong motion of heads will result in strong variation of light intensities reflection on faces. The strong variation introduced by motion are much stronger than the hidden weak heart rate signal in video. In this work, a framework is proposed to monitor the heart rate of human in intensive exercise with video camera. The Photoplethysmography (PPG) signal extracted from video is decomposed by singular spectrum analysis (SSA). And the motion information calculated from video is utilized to eliminate the frequency components caused by motion. Experimental results show that the proposed method could monitor human's heart rate in exercises accurately.