基于内容的图像特征相关性检索方法

Relevance retrieval scheme of image features in content-based image retrieval systems

针对目前基于内容的图像检索系统在模型优化及通用性设计等方面的局限性 ,以图像单一特征描述和提取方法为基础 ,根据人机交互的相关反馈原理 ,分析图像组合特征归一化权值系数 ,提出一套图像特征相关性检索新方案 ,并在着重对基于内容的图像特征相关性检索算法研究基础上 ,参照国际标准MPEG 7所提供的通用性系统设计规范 ,对新方案下的图像检索通用模型进行概念化设计 ,为解决基于内容的图像检索技术在Internet上实用化的瓶颈问题 ,提供一条新的思路 .

Aimed at the defects existed in content based image retrieval systems for optimization model and generalization design, a new scheme on relevance retrieval of image features in the systems is provided. Based on the method of describing and extracting single feature of image, the scheme analyzes unitary weight coefficient of assorted features of image and adopts interactive relevance feedback principle between man and computer. Under the research of relevance retrieval algorithm for image features and referring to the universal systems design norm provided in the MPEG 7, a conceptual design for optimizing the universal images retrieval model in the scheme is implemented. It provides a new idea for resolving the bottleneck problem: the practicability of content based image retrieval techniques on Internet. The rationality of the new scheme is validated by experimental results.