发明名称 Image Text Search and Retrieval System
摘要 Glyphs, symbols, or non-symbols are selected and become a “search term.” The search term is normalized and then depicted upon a grid for marking. Marked search term areas may be measured from a center point. The marks and measurements can be converted in to a distance map. For reference material loaded into a master database, each distance map may be given a unique code, which may correspond to a Unicode value. An end user may find new material to search, select a new search term and enter the search term into a disclosed system. The target term can be converted into distance maps and then matched with the distance maps of the master database. The pixels or extrapolated geometry of the search term can be used to match system documents that do not have an associated code.
申请公布号 US2015213593(A1) 申请公布日期 2015.07.30
申请号 US201514605886 申请日期 2015.01.26
申请人 Kim Sang Hun;de Fremery Peter Wayne 发明人 Kim Sang Hun;de Fremery Peter Wayne
分类号 G06T7/00;G06K9/00 主分类号 G06T7/00
代理机构 代理人
主权项 1. A system that can identify unencoded textual information in images and create a master database of glyphs that can be used to search the master database with target selections of glyphs, and return portions of the master database containing glyphs, the system comprising: a) a computer system comprising a processor, non-volatile memory, machine readable media, machine instructions upon the machine readable media, the master database, a user interface, an image processing module, an image matching module, a crude matching module, a coarse matching module, a fine matching module and an output module; b) the image processing module configured to accept selections of glyphs, match the glyphs, with like glyphs, convert the glyphs to bitmap images, map the glyphs upon grids, mark sections of the glyphs and create distance maps by marking a string of linear boxes with marks corresponding to the distance of the glyph marked sections to a glyph center point; c) the matching module configured to assign a value that may be Unicode to each glyph, and each distance map to create a database structure and database that is entered into the master database; d) the image processing module further configured to accept target selections of glyphs and to create corresponding distance maps as described in b) above; e) the image matching module configured match distance maps of target selections to distance maps of the master database; and f) the module of output configured to display sections of the master database containing target glyphs g) a coarse matching module configured to accept glyphs and enter glyph pattern data into the master database and the coarse matching module configured to match searched glyphs to glyphs of the master database by use of pattern matching.
地址 Seoul KR
您可能感兴趣的专利