基于超图形XGML的图像半结构化研究

项目名称： 基于超图形XGML的图像半结构化研究

项目编号： No.61271369

项目类型： 面上项目

立项/批准年度： 2013

项目学科： 无线电电子学、电信技术

项目作者： 袁家政

作者单位： 北京联合大学

项目金额： 82万元

中文摘要： 基于SVG的图像半结构化处理具有广泛的应用前景，已成为研究的一个热点。针对SVG以非结构化方式描述图像的局限性，利用图像整体与局部、局部与局部相似性，研究大容量古建筑图片和粗纹理文物工艺品图片中的各个局部实体对象及关系，以SVG指令集与图像描述的所有标记为基础，建立XGML的指令集与体系结构,该体系结构能够很好的描述半结构化图像和分形图形的功能；以经验模态分解EMD方法和Lagrange multiplier乘子为基础，研究一种多层次自适应分解的图像半结构化分离算法，分离二进制光栅图像转换为XGML的半结构化文档；与此同时研究一种基于支撑向量机SVM的XGML文档优化与压缩算法，以得到存储空间较少基于XGML半结构化图像文档。研究成果将解决基于Internet的虚拟数字化博物馆导航平台快速载入、展示大容量文物图像和基于文物图像内容的定位和检索所涉及的关键问题。

中文关键词： XGML；半结构化；图像分割；EMD；区域关系

英文摘要： There is a wide range of applications for images semi-structured processing based on SVG, it has become a hot research topic. There are limitations for the SVG unstructured way to describe the image, According to the overall image characteristics and local characteristics, the similarity between local features and local features, we study the various local entities objects and relationships in the capacity of the ancient architecture picture and coarse texture heritage crafts pictures. A XGML the instruction set architecture is proposed,it can be a good description of the function of semi-structured image and fractal graphics capabilities.Based on empirical mode decomposition method and Lagrange multiplier multiplier, a multi-level adaptive decomposition of the image semi-structured separation algorithm is designed, the raster image is separated and converted to the XGML the semi-structured documents. Meanwhile in order to get the image of storage space less documentation of Semi-structured document based on XGML,an XGML document optimization and compression algorithms based on SVM is proposed.The research will solve the key issues such as the Virtual digital museums navigation platform on Internt to quickly load and display the cultural relics image of the large-capacity, heritage image content-based locate an

英文关键词： XGML；Semi-structured；Image segmentation；EMD；Regional relations