Fernández-Caballero Antonio, Sokolova Marina V, Serrano-Cuerda Juan
Instituto de Investigación en Informática de Albacete (I3A), 02071 Albacete, Spain ; Departamento de Sistemas Informáticos, Universidad de Castilla-La Mancha, 02071 Albacete, Spain.
ScientificWorldJournal. 2013 Oct 31;2013:935026. doi: 10.1155/2013/935026. eCollection 2013.
Fall detection is an emergent problem in pattern recognition. In this paper, a novel approach which enables to identify a type of a fall and reconstruct its characteristics is presented. The features detected include the position previous to a fall, the direction and velocity of a fall, and the postfall inactivity. Video sequences containing a possible fall are analysed image by image using the lateral inhibition in accumulative computation method. With this aim, the region of interest of human figures is examined in each image, and geometrical and kinematic characteristics for the sequence are calculated. The approach is valid in colour and in infrared video.
跌倒检测是模式识别中的一个新出现的问题。本文提出了一种新颖的方法，该方法能够识别跌倒的类型并重建其特征。检测到的特征包括跌倒前的位置、跌倒的方向和速度以及跌倒后的静止状态。使用累积计算方法中的侧向抑制对包含可能跌倒情况的视频序列逐图像进行分析。为此，在每个图像中检查人体的感兴趣区域，并计算该序列的几何和运动学特征。该方法在彩色和红外视频中均有效。
