首页> 外国专利> Data indexing and similar vector searching method in high dimensional vector set based on hierarchical bitmap indexing for multimedia database

Data indexing and similar vector searching method in high dimensional vector set based on hierarchical bitmap indexing for multimedia database

机译:基于分层位图索引的多媒体数据库高维向量集中数据索引及相似向量搜索方法

摘要

Disclosed are a method of data indexing and pseudo vector search in a high-dimensional vector set based on a hierarchical bitmap index in a multimedia database. In the multimedia database according to the present invention, a method of indexing data and searching similar vectors in a high-dimensional vector set based on a hierarchical bitmap index is provided. A method of indexing a data set so that a short distance vector can be found in a short time and a similar vector search method using the same, the method comprising: (a) Bitmap by generating a bitmap for vectors in a multimedia database composed of a high-dimensional vector set; (B) performing an XOR operation between the bitmap indexes generated in step (a), and comparing the number of "11" and the difference between the upper and lower intervals of the bitmap index in the XOR operation result. Calculating by multiplying and excluding vectors whose approximate distance from the query vector is greater than a predetermined threshold value as vectors having significantly less similarity, and (c) vector and query vector not excluded in step (b). of Computing the distance and finally selecting as a vector similar to the query vector. According to the present invention, it is possible to index using a bitmap in a high-dimensional vector set, and in such an indexed state, it is possible to search the result vector set at an improved speed compared to the conventional method in searching for a similar vector similar to a query vector.;Multimedia Database, Hierarchical Bitmap Indexing, Data Indexing, Similar Vector Search
机译:公开了一种基于多媒体数据库中的分层位图索引在高维向量集中进行数据索引和伪向量搜索的方法。在根据本发明的多媒体数据库中,提供了一种基于分层位图索引在高维向量集中索引数据并搜索相似向量的方法。一种为数据集建立索引以便可以在短时间内找到短距离矢量的方法,以及使用该方法的类似矢量搜索方法,该方法包括:(a)通过在由以下组成的多媒体数据库中为矢量生成位图的位图高维向量集; (B)在步骤(a)中生成的位图索引之间执行XOR运算,并比较“ 11”的数目和XOR运算结果中位图索引的上下间隔之间的差。通过将与查询向量的近似距离大于预定阈值的向量相乘并排除为相似性明显较小的向量,以及将步骤c(b)中未排除的向量和查询向量相乘,来进行计算。计算距离并最终选择类似于查询向量的向量。根据本发明,可以在高维向量集中使用位图进行索引,并且在这种索引状态下,与传统的搜索方法相比,可以以更高的速度搜索结果向量集。类似于查询向量的相似向量。;多媒体数据库,分层位图索引,数据索引,相似向量搜索

著录项

  • 公开/公告号KR100786675B1

    专利类型

  • 公开/公告日2007-12-21

    原文格式PDF

  • 申请/专利权人

    申请/专利号KR20060019260

  • 申请日2006-02-28

  • 分类号H04N7/24;

  • 国家 KR

  • 入库时间 2022-08-21 19:54:29

相似文献

  • 专利
  • 外文文献
  • 中文文献
获取专利

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号