Human localization and identification by combining visual detection and UWB

This paper proposes a method to position and identify human in static scenes. The proposed framework comprises by the following parts: human detection, 2D localization calculating and visual-UWB sensing association. During the first stage, background segmentation is performed and then the segmented blobs would be filtered by the classification step performed by an artificial neural network. The second stage is related to scene reconstruction of the human positions from the images provided by several visual sensors used in the acquisition. The last procedure is to associate the detection results and human positions with the UWB positioning results.