Patent US5574802A successfully refreshed from Google Patents.
US5574802A
Expired / Lapsed
US
⏱️ 0.0 years remaining
Method and apparatus for document element classification by analysis of major white region geometry
Assignee: Fujifilm Business Innovation Corp
•
Filing: 1994-09-30
•
Granted: 1996-11-12
•
Expires: 2014-09-30
Abstract
Abstract A system for logically identifying document elements in a document image using structural models includes assistant function for users to describe the structural models, a geometric relationship comparator for comparing geometric relationships in a document to the geometric relationships in a structural model, a matching system for determining which structural model matches the document image, and a logical tag assigning system for assigning logical tags to the document elements of the document image based on the matching structural model. The method for logically identifying the document elements in the document image using structural models comprises identifying the major white regions in the document image, using geometric relationships between the major white regions in the document image, identifying a structural model which matches the document image, and assigning logical tags to the document elements in the document image based on the matching structural model.
Inventors
- 👤 Masaharu Ozaki
Bibliographic Details
Application Number
—
Jurisdiction
US
Data Source
google_patents
Last Synchronized
2026-10-30 22:50
CPC Classifications
G; G06; G06V; G06V30/00; G06V30/40; G06V30/41; G06V30/414; G06V30/416
Claims Analysis
Total Claims: 22
Claim 1
Independent Claim
238 words
A method for logically identifying document elements in a document image, comprising the steps of: storing a set of at least one predetermined structural model, each structural model in the set defining relationships between document elements of a corresponding model document; identifying major background regions in the document image: defining relationships between the major background regions in the document image; converting the defined relationships of the major background regions in the document image into nested one-dimensional relationships by selecting coextensive major background regions extending perpendicular to a direction of each level of the nested one-dimensional relationships; hierarchically comparing the nested one-dimensional relationships of the document image with the relationships of each structural model in the set; determining if the nested one-dimensional relationships of the document image match the relationships of at least one structural model in the set; selecting at least one matching structural model if the nested one-dimensional relationships of the document image match the relationship of the at least one structural model in the set to form a matching set of at least one structural model; identifying a most similar matching structural model from the matching set most similar to the relationships in the document image; and assigning appropriate logical tags to identify at least one of the document elements between the major background regions of the document image based on predetermined logical tags assigned to the document elements of the most similar matching structural model.
Claim 2
Independent Claim
238 words
A method for logically identifying document elements in a document image, comprising the steps of: storing a set of at least one predetermined structural model, each structural model in the set defining relationships between document elements of a corresponding model document; identifying major background regions in the document image: defining relationships between the major background regions in the document image; converting the defined relationships of the major background regions in the document image into nested one-dimensional relationships by selecting coextensive major background regions extending perpendicular to a direction of each level of the nested one-dimensional relationships; hierarchically comparing the nested one-dimensional relationships of the document image with the relationships of each structural model in the set; determining if the nested one-dimensional relationships of the document image match the relationships of at least one structural model in the set; selecting at least one matching structural model if the nested one-dimensional relationships of the document image match the relationship of the at least one structural model in the set to form a matching set of at least one structural model; identifying a most similar matching structural model from the matching set most similar to the relationships in the document image; and assigning appropriate logical tags to identify at least one of the document elements between the major background regions of the document image based on predetermined logical tags assigned to the document elements of the most similar matching structural model.
Claim 3
Dependent Claim
41 words
The method of claim 1, wherein the step of defining relationships between the element areas in the document image comprises the step of identifying a major relationship between at least one pair of element areas lying between the major background regions.
Claim 4
Dependent Claim
22 words
The method of claim 2, wherein the at least one major relationship comprises one of an above-below relationship and a left-right relationship.
Claim 5
Dependent Claim
34 words
The method of claim 1, wherein the step of defining relationships between element areas in the document image comprises the step of identifying minor relationships between the areas lying between the major background regions.
Claim 6
Dependent Claim
19 words
The method of claim 4, wherein the minor relationships comprise at least one of left-justified, right-justified, centered, and indented.
Claim 7
Dependent Claim
27 words
The method of claim 1, wherein the geometric relationships of the document image and the structural models are each defined as a nested hierarchy of one-dimensional relationships.
Claim 8
Dependent Claim
15 words
The method of claim 1, wherein the relationships define sizes of the major background regions.
Claim 9
Independent Claim
353 words
A method of logically identifying document elements in a document image, comprising the steps of: storing a set of at least one predetermined structural model, each structural model defining at least one level of a nested one-dimensional hierarchy of relationships defining separations between document elements of a corresponding model document; identifying major background regions in the document image; defining relationships between the identified major background regions of the document image; converting the defined relationships of the document image into at least one level of a nested one-dimensional hierarchy of relationships by selecting coextensive major background regions extending perpendicular to a direction of each of the at least one level of the nested one-dimensional hierarchy of relationships; selecting a top level of relationships of the at least one level of relationships of the document image as a current document image level; selecting a top level of relationships of the at least one level of relationships for the at least one stored structural model to form a set of current structural model levels; comparing the current document image level to each current structural model level of the set of current structural model levels; selecting each structural model whose current level matches the current document image level to form a set of matching structural models; determining if any additional document image levels exist; if so, selecting a next document image level as the current document image level and selecting a corresponding next level for each of the set of matching structural models as the current structural model level to reform the set of current structural models; repeating the comparing step through the next document image level selecting step until no additional document image levels exist; determining if a single selected structural model exists or a plurality of selected structural models exist; if a plurality of selected structural models exists, selecting a most similar one of the selected structural models that is most similar to the defined relationships of the document image as the single selected structural model; and identifying the document elements of the document image based on logical tags of the single selected structural model.
Claim 10
Independent Claim
353 words
A method of logically identifying document elements in a document image, comprising the steps of: storing a set of at least one predetermined structural model, each structural model defining at least one level of a nested one-dimensional hierarchy of relationships defining separations between document elements of a corresponding model document; identifying major background regions in the document image; defining relationships between the identified major background regions of the document image; converting the defined relationships of the document image into at least one level of a nested one-dimensional hierarchy of relationships by selecting coextensive major background regions extending perpendicular to a direction of each of the at least one level of the nested one-dimensional hierarchy of relationships; selecting a top level of relationships of the at least one level of relationships of the document image as a current document image level; selecting a top level of relationships of the at least one level of relationships for the at least one stored structural model to form a set of current structural model levels; comparing the current document image level to each current structural model level of the set of current structural model levels; selecting each structural model whose current level matches the current document image level to form a set of matching structural models; determining if any additional document image levels exist; if so, selecting a next document image level as the current document image level and selecting a corresponding next level for each of the set of matching structural models as the current structural model level to reform the set of current structural models; repeating the comparing step through the next document image level selecting step until no additional document image levels exist; determining if a single selected structural model exists or a plurality of selected structural models exist; if a plurality of selected structural models exists, selecting a most similar one of the selected structural models that is most similar to the defined relationships of the document image as the single selected structural model; and identifying the document elements of the document image based on logical tags of the single selected structural model.
Claim 11
Dependent Claim
43 words
The method of claim 8, wherein the step of defining relationships between the major background regions in the document image comprises the step of identifying a major relationship between at least one pair of document element areas lying between the major background regions.
Claim 12
Dependent Claim
22 words
The method of claim 9, wherein the at least one major relationship comprises one of an above-below relationship and a left-right relationship.
Claim 13
Dependent Claim
38 words
The method of claim 8, wherein the step of defining relationships between the major background regions in the document image comprises the step of identifying minor relationships between the document element areas lying between the major background regions.
Claim 14
Dependent Claim
27 words
The method of claim 11, wherein the minor relationships form a bottom most level of each structural model and a bottom most level of the document image.
Claim 15
Dependent Claim
19 words
The method of claim 11, wherein the minor relationships comprise at least one of left-justified, right-justified, centered, and indented.
Claim 16
Dependent Claim
36 words
The method of claim 8, further comprising the steps of: determining, after the structural model selecting step, if no structural models were selected; and if so, using the document image to form a new structural model.
Claim 17
Dependent Claim
23 words
The method of claim 8, wherein the storing step further comprises storing a regular expression for each level of each stored structural model.
Claim 18
Dependent Claim
22 words
The method of claim 15, wherein the converting step further comprises generating a document expression for each level of the document image.
Claim 19
Dependent Claim
36 words
The method of claim 16, wherein the comparing step comprises comparing the document expression corresponding to the current document image level to the regular expression corresponding to the current structural model level for each structural model.
Claim 20
Dependent Claim
121 words
The method of claim 8, wherein the storing step comprises the steps of: generating a top-level expression pattern of background regions for each of the at least one structural model based on corresponding structural model document image document elements; identifying pseudo-elements in the top-level expression pattern for each at least one structural model; generating a second or greater level expression pattern of background regions for each of the pseudo-elements in said each of the at least one structural model; identifying minor document element characteristics for each of the document elements of each of the at least one structural model; and generating minor relationships of background regions for each of the document elements of each of the at least one structural model.
Claim 21
Dependent Claim
199 words
A document element identifying system for logically identifying document elements of a document image, comprising: a memory for storing a plurality of stored structural models, each of the stored structural models comprising a regular expression; a document background region extraction system; major background region selecting means for selecting some major background regions extracted by the document background region extraction system from the document image and for generating a nested one-dimensional document expression for the selected major background regions wherein the selected major background regions coextensively extend perpendicular to a direction of each level of the nested one-dimensional document expression; regular expression matching means for hierarchically comparing the nested one-dimensional document expression to the regular expression of each of the plurality of stored structural models; matching means for identifying a single matching stored structural model based on the results of the comparison by the regular expression matching means; logical tag assigning means for assigning logical tags to identify the document elements of the document image based on the single matching stored structural model; and a processor controlling the document background region extraction system, the major background region selecting means, the logical tag assigning means, the matching means and the memory.
Claim 22
Dependent Claim
199 words
A document element identifying system for logically identifying document elements of a document image, comprising: a memory for storing a plurality of stored structural models, each of the stored structural models comprising a regular expression; a document background region extraction system; major background region selecting means for selecting some major background regions extracted by the document background region extraction system from the document image and for generating a nested one-dimensional document expression for the selected major background regions wherein the selected major background regions coextensively extend perpendicular to a direction of each level of the nested one-dimensional document expression; regular expression matching means for hierarchically comparing the nested one-dimensional document expression to the regular expression of each of the plurality of stored structural models; matching means for identifying a single matching stored structural model based on the results of the comparison by the regular expression matching means; logical tag assigning means for assigning logical tags to identify the document elements of the document image based on the single matching stored structural model; and a processor controlling the document background region extraction system, the major background region selecting means, the logical tag assigning means, the matching means and the memory.
References Cited (Backward Citations)
US4698779A
Cited
US4907285A
Cited
US4876728A
Cited
US5046114A
Cited
US4887302A
Cited
US5185813A
Cited
US5272764A
Cited
US5430808A
Cited
US5335298A
Cited
USPTO Annuity & Maintenance Windows
| Annuity Stage | Window Opens | Payment Deadline | Surcharge Period Until | Estimated Fee | Status |
|---|---|---|---|---|---|
| 3.5 Years (Year 4) | 1999-11-12 | 2000-05-13 | 2000-11-12 | $2000.00 | Window Closed / Lapsed |
| 7.5 Years (Year 8) | 2003-11-12 | 2004-05-13 | 2004-11-12 | $3760.00 | Window Closed / Lapsed |
| 11.5 Years (Year 12) | 2007-11-12 | 2008-05-13 | 2008-11-12 | $7700.00 | Window Closed / Lapsed |