Intelligent Human Fall Detection for Home Surveillance

We propose in this paper an intelligent system for human fall detection in indoor environments. It can serve as home surveillance for elderly person. Specifically, we detect the human body captured by camera using the Vibe algorithm. Then Gabor feature of a human body is extracted as observation feature. Based on the extracted feature, incidents are detected as the changes from the standing state to the fall state in the feature space. The feature is detected on single image and is effective and efficient. Compared with the motion features across images combined with biological inspired feed-forward network, our method can obtain more robust detection results.