Research on human body detection methods based on the head features on the disaster scenes

Human detection is a new rising research area, involving pattern recognition, image processing, computer vision, human kinematics, artificial intelligence and machine learning and so on. It is a multi-disciplinary research direction. Head can be recognized by computer easily. The first feature selected by a robust and efficient human body detection system must be head features. Nowadays, the human body detection based on head feature is widely used in intelligent control, human-computer dialog, pedestrian detection, disaster rescue and so on. Because of the complex background, complex lighting condition, the block and rotation of body and so on, we don´t select the overall characteristics of the human body. We select the head features which are not easily deformed and not easily sheltered. This paper proposed a human detection method of head feature based on skin color, Haar feature and HOG feature combined with color images and infrared images. The experiment shows that the detection method adapts to the needs of the disaster site and has good accuracy and can be real-time.