IDENTIFICATION METHOD FOR PART OF DOCUMENT IMAGE

摘要

PROBLEM TO BE SOLVED: To dynamically specify the feature of a document image inside a document corpus by using the attribute of a layout object and selecting it. SOLUTION: After the page image 226 of the document image is recorded, the image segmenter of a document corpus management/search system 140 divides the respective page images 226 into one or plural layout objects 238. Then, image attributes 240 corresponding to the respective divided layout objects are calculated. By using the attributes 240, the feature of the document image is defined. When the corpus is prepared (made into a population), for the respective page images 226, the selection routine of the feature is executed, a set of the layout objects is consumed and the new set of the layout objects is generated. The new set of the layout objects is recorded in a file system 117 as the feature 242 of the respective page images 226.