计算机应用 ›› 2016, Vol. 36 ›› Issue (8): 2274-2281.DOI: 10.11772/j.issn.1001-9081.2016.08.2274

• 人工智能 • 上一篇    下一篇

图像自动标注技术研究进展

刘梦迪1, 陈燕俐1, 陈蕾1,2,3   

  1. 1. 南京邮电大学 计算机学院, 南京 210003;
    2. 江苏省无线传感网高技术研究重点实验室(南京邮电大学), 南京 210003;
    3. 南京航空航天大学 计算机科学与技术学院, 南京 210016
  • 收稿日期:2015-12-14 修回日期:2016-03-21 出版日期:2016-08-10 发布日期:2016-08-10
  • 通讯作者: 刘梦迪
  • 作者简介:刘梦迪(1993-),女,河南通许人,硕士研究生,CCF会员,主要研究方向:大规模机器学习;陈燕俐(1969-),女,江苏苏州人,教授,博士,主要研究方向:智能信息处理、网络信息安全;陈蕾(1975-),男,江西宜春人,副教授,博士,CCF会员,主要研究方向:大规模机器学习、模式识别、数据挖掘。
  • 基金资助:
    国家自然科学基金资助项目(61572263);中国博士后科学基金资助项目(2015M581794);江苏省高校自然科学研究面上项目(15KJB520027);江苏省博士后科研资助计划项目(1501023C);南京邮电大学校级科研基金资助项目(NY214127,NY215096)。

Advances in automatic image annotation

LIU Mengdi1, CHEN Yanli1, CHEN Lei1,2,3   

  1. 1. College of Computer, Nanjing University of Posts and Telecommunications, Nanjing Jiangsu 210003, China;
    2. Jiangsu High Technology Research Key Laboratory for Wireless Sensor Networks(Nanjing University of Posts and Telecommunications), Nanjing Jiangsu 210003, China;
    3. College of Computer Science and Technology, Nanjing University of Aeronautics and Astronautics, Nanjing Jiangsu 210016, China
  • Received:2015-12-14 Revised:2016-03-21 Online:2016-08-10 Published:2016-08-10
  • Supported by:
    This work is partially supported by the Fundamental Research Funds for the Central Universities (JD1413).

摘要: 现有图像自动标注技术算法可以大致划分为基于语义的标注算法、基于矩阵分解的标注算法、基于概率的标注算法以及基于图学习的标注算法等四大类。介绍了各类别中具有代表性的图像自动标注算法,分析了这些算法的问题模型及其功能特点,并归纳了图像自动标注算法中主要的优化求解方法及算法评价中常用的图像数据集和性能评价指标。最后,指出了图像自动标注技术目前存在的主要问题,并且提出了这些问题的解决思路。分析结果表明,对于图像自动标注技术的研究,可充分利用现有算法的优势互补,或借助多学科交叉的优势,寻找更有效的算法。

关键词: 图像检索, 图像自动标注, 标签填补, 标签去噪, 标签预测

Abstract: Existing image annotation algorithms can be roughly divided into four categories:the semantics based methods, the probability based methods, the matrix decomposition based methods and the graph learning based methods. Some representative algorithms for every category were introduced and the problem models and characteristics of these algorithms were analyzed. Then the main optimization methods of these algorithms were induced, and the common image datasets and the evaluation metrics of these algorithms were introduced. Finally, the main problems of automatic image annotation were pointed out, and the solutions to these problems were put forward. The analytical results show that the full use of complementary advantages of the current algorithms, or taking multi-disciplinary advantages may provide more efficient algorithm for automatic image annotation.

Key words: image retrieval, automatic image annotation, tag completion, tag denoising, tag prediction

中图分类号: