计算机科学与探索 ›› 2015, Vol. 9 ›› Issue (3): 257-265.DOI: 10.3778/j.issn.1673-9418.1411002

• 综述·探索 • 上一篇    下一篇

图像目标类别检测综述

蔡  强1+,刘亚奇1,曹  健1,毛典辉1,牛  群2   

  1. 1. 北京工商大学 计算机与信息工程学院,北京 100048
    2. 中央民族大学 管理学院,北京 100081
  • 出版日期:2015-03-01 发布日期:2015-03-09

Review on Object Class Detection of Images

CAI Qiang1+, LIU Yaqi1, CAO Jian1, MAO Dianhui1, NIU Qun2   

  1. 1. School of Computer and Information Engineering, Beijing Technology and Business University, Beijing 100048, China
    2. School of Management, Minzu University of China, Beijing 100081, China
  • Online:2015-03-01 Published:2015-03-09

摘要: 随着移动设备与社交网络的迅速发展,数字图像的数据规模急剧增加,图像目标类别检测已经发展成为目前计算机视觉领域内的一个研究热点。对图像目标类别检测的关键问题进行了综述。首先对目标类别检测的研究背景进行了介绍;然后对目标类别检测技术进行了综述,其中包括外观模型、分类器和定位策略3个核心技术,以及数据集和评价标准;最后列出了目前目标类别检测算法的测试结果,并总结了目标类别检测的主要研究难点和发展方向。

关键词: 目标类别检测, 外观模型, 分类, 定位策略, 图像数据集, 评价标准

Abstract: With the development of mobile devices and social networking services, the data size of digital images has increased rapidly. Object class detection of images has become one of the most focused areas in computer vision in the new century. This paper attempts to make a review on the object class detection. At first, this paper makes a brief introduction about object class detection. Then, this paper provides a comprehensive survey from three aspects of object class detection: core techniques which include the appearance model, classification and location strategies, image datasets and evaluation methods. Finally, this paper lists the state-of-art results, and summarizes the challenge and future trend of object class detection.

Key words: object class detection, appearance model, classification, location strategies, image dataset, evaluation