基于t混合模型的自动图像标注

Automatic Image Annotation Based on t Mixture Model

图像自动标注的实质是通过对图像视觉特征的分析来提取高层语义关键词用于表示图像的含义,从而使得现有图像检索问题转化为技术已经相当成熟的文本检索问题,在一定程度上解决了基于内容图像检索中存在的语义鸿沟问题。采用t混合模型在已标注好的训练图像集上计算图像区域类与关键字的联合概率分布,在此基础上,对未曾观察过的测试图像集,利用生成的模型根据贝叶斯最小错误概率准则实现自动图像标注。实验结果表明,该方法能有效改善标注结果。

The essence of automatic image annotation is to obtain the semantic keywords of images from visual features and to support the semantic level search, then image retrieval can be transformed into text retrieval,which is fairly mature.To a certain extent,it can solve the semantic gap existing in content-based image retrieval.This paper was based on t mixture model,and computed a joint probability distribution for image regions classes and keywords.On this basis,it annotated unseen set of test images by used model according to Bayes minimum error probability criterion.The experiments results show that this method can significantly improve the labeling results.