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.