Age Classification of Evacuees at Times of Disaster Using a Vibration Sensor

In this study, a method to detect children evacuating from a building at the time of a disaster using a vibration sensor is proposed. The features for detection are obtained using linear prediction, which is suitable for analyzing resonance phenomena such as speech. Subsequently, the extracted features are used in the linear discrimination analysis to discriminate between children and adults. Experimental results with 40 subjects show that the detection rate was approximately 80% in the test environment.