Indoor Human Fall Detection Method Based on Infrared Images and Back-Projection Algorithm

Falls are reported to be a major cause of injury in China's elderly population. Shortening the time between the fall and subsequent treatment can reduce injuries caused by falls; therefore, the demand for indoor fall detection is increasing annually. Infrared image-based human fall detection methods are becoming increasingly popular owing to advantages such as being unaffected by light and non-intrusive. However, the traditional methods perform low accuracy because it is difficult to extract the features from the infrared videos at low resolution and high noise. Hence, this paper proposes an indoor human fall detection method based on a back-projection algorithm. First, the distance between the human body and the sensor is calculated using the human body temperature. Second, the height of the human body in the real world is reversely deduced using image information. Finally, the human body height data are smoothed and used for fall detection based on the height variation. The experimental results show that the detection accuracy of the proposed method is 98.57%, which is better than that of traditional projection methods. Therefore, it can be used for detection in real-life situations.