计算机工程与应用 ›› 2007, Vol. 43 ›› Issue (2): 53-53.

• 学术探讨 • 上一篇    下一篇

一种基于区域特征关联的图像语义标注方法

陈世亮,李战怀,袁柳   

  1. 西北工业大学计算机学院
  • 收稿日期:2006-05-19 修回日期:1900-01-01 出版日期:2007-01-11 发布日期:2007-01-11
  • 通讯作者: 陈世亮 chensl

A image semantic annotating method based on region features relevancy

ShiLiang Chen,,   

  1. 西北工业大学计算机学院
  • Received:2006-05-19 Revised:1900-01-01 Online:2007-01-11 Published:2007-01-11
  • Contact: ShiLiang Chen

摘要: 图像语义的标注需要解决图像高层语义和底层特征间存在的语义鸿沟。采用基于图像分割、并结合图像区域特征抽取的方法,建立图像区域语义与底层特征间的关联,采用基于距离的分类算法,计算区域特征间的相似性,并对具有相同或相近特征的区域的语义采用关联关键字的方法进行区分,用关键字实现图像语义的自动标注。

关键词: 图像标注, 区域语义, 关联关键字

Abstract: The semantic gap between image semantic and visual features should be solved in image annotation. Through the image segmenting and visual content extracting of regions, the relationships between image semantic of regions and visual features are established. The classification algorithm based on distance is used to measure the distance between regions, co-occurrence keywords are adopted to discriminate regions which have identical or similar features, then the image semantics is annotated by keywords automatically.

Key words: image annotation, region semantic, co-occurrence keywords