Title :
Color spaces and histogram distances influence on people re-identification in video stream
Author_Institution :
VGTU, Vilnius, Lithuania
Abstract :
The influence of color spaces and histogram distances on object re-identification in video stream is investigated. Dataset with images of pedestrians is being analyzed. A set of color spaces, histograms types and distance functions is applied to minimize re-identification errors by calculating false acceptance and false rejection error rates.
Keywords :
image colour analysis; object detection; pedestrians; video streaming; video surveillance; color spaces; false acceptance error rate; false rejection error rate; histogram distance function; histogram types; pedestrian images; people re-identification; re-identification error minimization; video stream; Cities and towns; Error analysis; Histograms; Image color analysis; Noise; Streaming media; Tracking; color space; distance function; histogram; re-identification; tracking;
Conference_Titel :
Electrical, Electronic and Information Sciences (eStream), 2015 Open Conference of
Conference_Location :
Vilnius
DOI :
10.1109/eStream.2015.7119489
