摘要
基于内容的图像搜索是目前搜索领域的研究热点,取得了较大的成果,在实际的开发中对于图像的数据库存储和检索的要求较高。本文通过对图像特征的研究,采用了面向对象的多层数据模型,并分析了索引空间中的多维索引降维、向量空间、度量空间等方法,引出了基于预分类和预计算相似度、树索引和二分快速检索图像数据的方法。接着通过分析其工作流程,从数据模型和检索方法上进行了论述,构造了基于内容的图像检索的工作架构。本文总结了此方案的创新点和适用范围,从而在理论证明了它的优秀的检索性能。最终提出了采用此方案进行基于内容的图像查询的可行途径。
CBIR （Content-based image retrieval） is the hot field of search, now it takes much progress. But in the actual development, the requirements for storage and retrieval of image database are higher. So in this paper, we explain image feature, use of object oriented data model. Then, we discuss the Dimensionality Reduction, SAM, MAM in the index space. And, a way basing on Preclassification, Similarity Precomputation, Tree index and Binary search is proposed. This paper analyzes the workflow, data model and the retrieval method, then put forward a working body. At last, we summarize the point of innovation and applicability, theoretically prove that the superior retrieval performance.
出处
《价值工程》
2012年第8期114-116,共3页
Value Engineering
