摘要: 为提高图像标注质量,提出一种反馈日志与混合概率模型相结合的图像标注方法。利用本体语义网计算标注词之间的相似性度,将相似度应用于日志分析,得到具体应用中的标注词间关系,结合标注词间的关系和图像底层特征,使用混合概率模型进行自动图像标注。实验结果表明,该方法能获得较好的查全率和查准率。
关键词:
图像标注,
语义鸿沟,
混合概率模型,
日志,
语义相关度
Abstract: To improve the quality of automatic image annotation. A image annotation method combined of feedback log and mixture probabilistic model is proposed in this paper. It calculates the similarity of words by using ontology semantic-web, analyzes the feedback logs by using these similarity and gets the correlation of words in this application, combines the correlation and low-level image features to annotation images. Experimental results show that this method can achieve good recall ratio and precision ratio.
Key words:
image annotation,
semantic gap,
mixture probabilistic model,
log,
semantic similarity
中图分类号:
黄勇辉, 尚赵伟, 张明新. 反馈日志与混合概率模型相结合的图像标注[J]. 计算机工程, 2012, 38(21): 202-205.
HUANG Yong-Hui, CHANG Diao-Wei, ZHANG Meng-Xin. Image Annotation Combining Feedback Log and Mixture Probabilistic Model[J]. Computer Engineering, 2012, 38(21): 202-205.