图像自动标注方法研究综述

Summary of Automatic Image Annotation Method

随着Web2.0的逐步发展,海量用户生成的图像信息充斥于各大网络平台,图像自动标注技术逐步成为图像检索以及图像理解的关键问题之一。该文主要通过对现有图像自动标注方法的文献进行收集和整理,在比较、分析各种方法相关理论和实现技术的基础上,对图像自动标注方法研究进展进行评述;并归纳了各种方法的优势与不足。得出结论:图像自动标注方法和图像处理技术仍然需要从机器学习方面进一步的研究与改进,且可以从图像信息的标注拓展到视频信息的标注。

With the progressive development of Web2. 0,massive user- generated image informa- tion filled in every network platform,automatic image annotation technology gradually become one of key issues of the image retrieval and image understanding. In this paper,through collecting and organizing documents of the existing automatic image annotation method to understand the theory and analysis of the various methods,On this basis,the status of Automatic Tagging images are reviewed,and comparative analysis of the advantages of each method and insufficient. The conclusion is： automatic image annotation method and image processing technology still needs further research and improvement from the active learning,and can expand the image information from the label to label video information.