基于意象认知模型的图像检索技术  

Image retrieval technology based on imagery cognition model

作　　者：谭征宇[1] 孙守迁[1] 

机构地区：[1]浙江大学计算机科学与技术学院,浙江杭州310027

出　　处：《浙江大学学报（工学版）》2008年第5期763-767,842,共6页Journal of Zhejiang University：Engineering Science

基　　金：国家自然科学基金资助项目(60475025);国家教育部博士点基金资助项目(20050335096)

摘　　要：鉴于在产品造型设计过程中设计师对于图像检索技术的需求,提出了塑料产品的意象认知模型,并构建了基于该模型的意象看板系统(E-image board system).该系统运用语意差异法分别建立了塑料椅子的形态和材质意象认知模型,利用距离分布直方图(DDH)结合Hu不变矩的形状提取与表示算法,提取和表示产品图像中的形态特征,采用纹理谱表示方法提取和表示产品图像中的纹理特征,并运用相似性度量和图像意象的相关度标注技术进行系统数据库的扩充和标注.该技术克服了传统检索方式不能体现用户意象的问题,满足了设计师利用意象词汇检索符合用户意象认知模型的产品图像的需求.An imagery cognition model of plastic products was advanced to satisfy the industrial designers’ demands for image retrieval in product design, based on which an E-image board system was designed. The shape and material imagery models of plastic chairs were built by semantic differential method. The distance distribution histogram （DDH） and Hu fixedness matrix were used to extract and express the products’ image shape characteristics. And the texture characteristics were obtained and described by texture chart. Comparability measurement and imagery correlation label techniques were combined to expand and label the system database. This technique satisfies the designers’ requirements to retrieve product image by imagery words in consistence with users’ imagery cognition model, and it conquers the shortcoming in the traditional retrieval model which cannot describe users＇ imagery.

关 键 词：意象认知 塑料产品 图像检索 

分 类 号：TB472[一般工业技术—工业设计] TN911.73[电子电信—通信与信息系统]