Pre collision detection system for pedestrian safety based on HOL

Pedestrian detection is an important task in intelligent video surveillance systems. In this work, we propose a system aimed to detect pedestrians from images. Our approach begins by extracting feature´s object and classify them by a Support Vector Machine in order to detect the presence of a pedestrian. We adopt a new image descriptor called Histograms of Oriented Laplacian (HOL), which extract the most relevant image features. The method is developed and validated with the experimental results to evaluate the approach´s efficiency against other descriptors.