机构地区: 广州大学华软软件学院网络技术系
出 处: 《软件》 2016年第4期40-45,共6页
摘 要: 由于其实用价值和理论价值,目标检测是智能视频监控技术研究的重点,也是计算机视觉领域的一个研究热点,引起了研究者广泛关注。本文根据视频图像背景和前景目标的动或静的情况进行分类,将目标检测问题分为基于背景建模的目标检测和基于目标建模的目标检测两类。对于每类问题,分别全面综述了该问题的发展、常用算法模型及当前的研究成果等,然后讨论了对各类算法模型的评测指标、评测数据集和评测结果,最后总结了当前这两类目标检测方法存在的不足以及给出了对未来发展的思考和展望。 Because of its practical and theoretical values, the object detection has become a focus of research on intelligent video surveillance, and it has become a hot topic in the areas of computer vision, attracting many researchers’ attentions. This paper categorizes the video-based object detection according to the video frames' background and fore-ground conditions into two main categories: background modeling and object modeling. For each category, the devel-opment history, common algorithm model and state-of-the-art research achievements are reviewed comprehensively. Afterwards, we discuss the evaluation index, evaluation dataset and evaluation results of different object detection algo-rithms. At the end of this paper, shortcomings of these two kinds of object detection methods are summarized; thoughts and foresights of this field are given.
领 域: [自动化与计算机技术] [自动化与计算机技术]