应用Web标注技术的建筑图像语义采集方法

发布时间：2018-08-26 13:39

【摘要】：为解决建筑师难以快速地从互联网中检索到符合创作需求的建筑图像的问题,提出了应用Web标注技术的建筑图像语义采集方法.首先,从建筑学角度界定了建筑图像及建筑图像语义的概念和类型;其次,给出了该方法的总体框架和操作流程;最后,以著名建筑网站为例进行案例演示,验证了该方法的可行性和有效性.操作流程细分为3个步骤,以人工添加和在线学习的方式建立建筑语义词典;运用数据采集软件,从建筑图像所在网页中分别采集图像名称、图像注释、图像周围文本、所在网页标题、所在网页正文、图像超链接网页标题6项图像相关文本;根据图像语义提取规则,从上述文本中提取建筑图像语义,与图像文件建立关联后存储到建筑图像数据库.案例检验结果表明,该方法是可行的,具有较强的操作性,能够自动、批量地从互联网中下载建筑图像,并采集图像名称、图像类别、图像主题、项目名称、项目类型等30多项特征,有效地克服了建筑图像查询效率较低的问题,进而提升了建筑师运用互联网图像进行创作的能力.
[Abstract]:In order to solve the problem that it is difficult for architects to retrieve architectural images from the Internet quickly, a semantic acquisition method of architectural images based on Web tagging technology is proposed. Firstly, the concept and types of architectural image and architectural image semantics are defined from the architectural point of view. Secondly, the overall framework and operation flow of this method are given. Finally, a case study of famous construction website is presented. The feasibility and effectiveness of the method are verified. The operation flow is subdivided into three steps to establish the architectural semantic dictionary by manual addition and online learning, and to collect the image name, image annotation and the text around the image from the web page of the building image by using the data acquisition software. The title of the web page, the body of the web page, the image hyperlink page title, 6 items of image related text, and the architectural image semantics are extracted from the text according to the rules of image semantic extraction. After establishing an association with the image file, it is stored in the building image database. The results of case study show that the method is feasible and has strong maneuverability. It can automatically and batch download building images from the Internet, and collect image names, image categories, image themes, project names, etc. More than 30 features, such as project types, effectively overcome the problem of low efficiency of architectural image query, and then enhance the ability of architects to create images on the Internet.
【作者单位】：哈尔滨工业大学建筑学院;哈尔滨工业大学建筑计划与设计研究所;黑龙江省寒地建筑科学重点实验室(哈尔滨工业大学);
【基金】：国家自然科学基金(51178132)
【分类号】：TP391.41;TU204

【相似文献】