| 摘要 |
PROBLEM TO BE SOLVED: To search a document image at high speed by using only appropriate symbol partial sets from symbol strings even when there are many document images to be searched without using a device which requires additional information like an OCR. SOLUTION: A symbol string searching means 1 extracts a prescribed length of partial string, while sequentially deviating the input symbol string one by one from the top. Whenever the partial string is extracted, it is checked whether a document frequency, which is obtained from a symbol string indexing means 4, satisfies a predetermined condition or not, and then, partial strings satisfying the condition are collected as those to be used for search. When the collected partial strings are used, search takes place in the same procedure as that of a method for synthesizing search with the use of a plurality of search words in known text search. COPYRIGHT: (C)2009,JPO&INPIT
|
