Image understanding system
AssigneeHitachi, Ltd.
InventorsYasuaki Nakano, Hiromichi Fujisawa, Junichi Higashino, Masakazu Ejiri
Life of this patent
Estimating the term…
Abstract
An image understanding system of this invention uses a grammer describing a document image, and represents the structure of an unknown input image by parsing a statement (the structure of the grammar) written in accordance with this grammer. In other words, the grammer describes an image as substructures and the relative relation between them, and when the substructures and their relative relation are identified in parsing, search is then made whether or not the substructures and their relative relation exist in an unknown input image. The structure of the unknown input image is represented on the basis of the result of this search.
Claims12
A system for automatically understanding structure of an image provided by a document having an unknown structure comprising:
- means for optically scanning an input image and for converting said input image to a digitized image by photoelectric conversion;
- means for storing said digitized image;
- means for storing statements written in accordance with a grammar describing hierarchically an image structure of a document as a group of substructures by means of a plurality of rules defining conditions under which respective substructures exist in a document and relative positional relation and clearance between said substructures on the document, wherein each of said substructures has a format label of the document, and wherein the image structure of the document is described in a general form;
- means for reading out sequentially said statements of a document from said means for storing statements and for searching whether any of a plurality of regions of said document satisfy any one of said conditions to thereby indicate the existence of a substructure in said input image;
- means for identifying actual regions corresponding to substructures indicated as existing in said input image on the basis of a result of operation of said means for reading and searching and for classifying said identified actual regions in accordance with respective format labels of said existing substructures.
A system for automatically understanding according to claim, wherein said substructures and their relative positional relation are described by operators designating characteristic features of the document.
A system for automatically understanding structure of an image provided by a document having an unknown structure, comprising:
- means for storing statements written in accordance with a grammar describing a generic layout structure of a document as a group of rectangular regions where each region can be recursively defined in terms of smaller regions, each of said regions having a format label of the document and said statements include conditions for defining when respective regions exist in the document, said conditions are satisfied by searching the document, wherein the image structure of the document is described in a generic form;
- means for converting an input image to a digitized image by photoelectric conversion;
- means for storing said digitized image;
- means for extracting rectangular regions from said digitized image;
- means for matching said statements stored in said means for storing with said rectangular regions extracted from said digitized image according to said conditions and for identifying a location of each rectangular region in the document described in a generic form;
- and
- means for classifying said identified rectangular regions corresponding to a format label thereof and for recognizing characters within the regions.
Dependencies are inferred from the claim text.