Abstract: The target of image annotation is to affine text information to corresponding images，which helps a lot in managing and retrieving large collections of images.Although it has been studied for several years，it’s still a challenging research area in computer vision and machine learning.Various of methods have been utilized to annotate unlabeled images.This paper reviews the state of the art algorithms and models for annotating images with keywords，including traditional classification methods，relevance models，topic models，utilizing context information with random fields and annotation with the help of Internet.Several challenges of image annotation are also discussed in this paper.

Key words: image annotation, computer vision, machine learning

摘要： 图像标注的目标是针对每幅图像，利用相对应的文本信息进行描述，从而能够对海量的图像数据进行有效的管理和检索。尽管图像标注已经被研究了若干年，然而它仍然是机器视觉和机器学习领域中一个非常具有挑战性的问题。各种各样的算法被用于图像的标注工作。对目前基于关键词的图像标注的一些常用的算法和模型进行了综述，包括传统的基于分类的方法、相关模型、主题模型、基于随机场的上下文信息的处理以及利用Internet上海量的数据来辅助图像标注等等。讨论了目前图像标注研究中遇到的一些具有挑战性的问题。

关键词: 图像标注, 机器视觉, 机器学习