Method of extracting characteristic facial expression for monitoring system

We propose a method of extracting characteristic facial expressions from video for monitoring system regarding the elderly. Over the past few years, the number of the elderly living alone has increased; therefore, various systems for monitoring the safety of the elderly have been developed and put into practice. There have been studies on monitoring systems using robots at elderly homes. Such systems are expected to be put into practice. Our proposed method extracts several characteristic facial expression images from video images to estimate the approximate psychological state of the elderly rather than conveying safety information, as with conventional methods. Specifically, facial expression images are extracted using an index of the difference in features between criteria facial image and facial images detected from each frame.