Detección y rastreo de peatones empleando vectores de características de histogramas de gradientes orientados y patrones binarios locales en una máquina de soporte vectorial

Abstract

The detection and identification of pedestrians, as well as their tracking, is an open problem in the area of artificial vision that has been worked in many different ways. This paper presents a proposal for the detection of pedestrians using image analysis techniques for the extraction of characteristics of a pedestrian based on histogram of oriented gradients and local binary patterns which are extracted from a region of interest previously detected, combined and entered to a vector support machine that functions as a classifier between samples of pedestrians and non-pedestrians. A tracker based on the MeanShift algorithm is implemented for its efficiency in tracking processing and identifying the new locations of each of the detected targets.