基于统计分类的行人检测方法综述

Survey on Pedestrian Detection Based on Statistical Classification

作为计算机视觉以及智能交通领域一个重要的研究方向,行人检测技术近年来受到业界广泛关注,其中基于统计分类的方法因为其检测效果好,得到了广泛应用。对基于统计分类的行人检测方法进行综述,归纳了目前用于行人检测的数据集,并给出了每个数据集中各种样本的分布情况;接着总结了行人检测方法中用于目标识别的行人特征及其改进方式;对行人检测主要分类器做了概述;指出了行人检测目前存在的问题并展望了未来的发展方向。

As one of the most important areas of research in the domain of computer vision and intelligent tranffic, pedestrian detection has attacted extensive interest from the research community in recent years. Because the detection based on statistical classification method is efficient, it has been widely applied. Pedestrian detection method was focused based on statistical classification. Standard dataset for evaluation of human detection was reviewed, studying the statistics about these dataset. Extracted features and the improvement on them were summarized. The main classifier for pedestrian detection was summarized. The present problems and future research trends in pedestrian detection were proposed.