Abstract: In the past few years, content-based image retrieval has been becoming an active research area. There exists a gap between high-level concepts and low-level features. Relevance feedback is a promising approach to finding a mapping between semantic objects and low-level features. Feature weighting is also an important issue of multiple features combination in content-based image retrieval. This paper proposes a feature-weighting scheme based on rough set in relevance feedback. During the feedback process, a decision table is constructed. Then the weight of a feature space is determined by the precision of the decision rules. The experiments show that this approach is effective in feature weighting for content-based image retrieval, which gets higher efficiency than Rui’s algorithm.

Key words: CBIR, Rough set, Relevance feedback

摘要： 为了提高图像检索的效率，近年来相关反馈机制被引入到基于内容的图像检索领域，而在基于内容的图像检索系统中，多特征融合检索中的特征加权又是一个重要的问题。该文提出了一种新的基于特征加权的相关反馈方法，在粗集理论的基础上，结合用户标记的反馈图像建立决策表，通过决策规则的精度来对多个特征加权，使图像检索和人的感知更加接近。实验表明该方法是有效的，并较Rui的相关反馈方法在性能上有很大提高。

关键词: CBIR, 粗糙集, 相关反馈

CLC Number: 

• TP391