Back to results
US 9,805,258 B2Granted
Systems and methods for separating documents
AssigneeKofax International Switzerland Sarl
InventorJason Michael Neavin
Life of this patent
Estimating the term…
Abstract
A method of automatically separating a stack of pages into documents includes receiving a plurality of page images from a scanner; extracting content from each of the plurality of page images; for each of the plurality of page images, determining, based upon the extracted content, whether the page image is an initial page image of a document; and appending each of the plurality of page images located between a first determined initial page image and a second determined initial page image to the first determined initial page image.
Claims18
A method of processing a plurality of page images belonging to a plurality of documents so as to separate, without the use of separator sheets, the page images into individual ones of the plurality of documents, the method comprising:
- receiving the plurality of page images from a scanner;
- extracting content from each of the plurality of page images;
- for each of the plurality of page images, determining, based upon the extracted content, whether the respective page image is an initial page image of one of the plurality of documents, wherein the determining comprises:
- assigning a classification weight to each of the plurality of page images based upon an analysis of the extracted content, wherein each classification weight reflects a discrete probability that the respective page image is one of the initial pages of one or more of the plurality of documents;
- and
- determining, based upon a comparison of the assigned classification weight to a sample set of classification weights, whether the page image is an initial page of a document in the plurality of documents;
- wherein the comparison comprises calculating a percentage probability of whether the page is an initial page of one of the plurality of documents according to a formula:
- (1−((Wtype X−WAssigned)/Wtype X));
- wherein Wtype X is a classification weight of a page of a document of a particular type in a sample set of documents;
- and
- wherein Wassigned is the classification weight assigned to the page image;
- and
- appending each of the plurality of page images located between a first determined initial page image of the one of the plurality of documents and a subsequent, second determined initial page image of another of the plurality of documents to the first determined initial page image.
10 more elements
A method of categorizing a plurality of scanned pages without the use of separator pages, the method comprising:
- extracting content from each of the plurality of scanned pages;
- assigning, based upon the extracted content, a classification weight to each of the plurality of scanned pages;
- determining for each of the plurality of scanned pages if, based upon the assigned classification weight, the scanned page is an initial page of a document, wherein the determining comprises:
- assigning a classification weight to each of the plurality of scanned pages based upon an analysis of the extracted content, wherein each classification weight reflects a discrete probability that the respective scanned page is the initial page of the document;
- and
- determining, based upon a comparison of the assigned classification weight to a sample set of classification weights, whether the scanned page is the initial page of a the document;
- wherein the comparison comprises calculating a percentage probability of whether the scanned page is the initial page of the document according to a formula:
- (1−((Wtype X−WAssigned)/Wtype X));
- wherein Wtype X is a classification weight of a page of a document of a particular type in a sample set of documents;
- and
- wherein Wassigned is the classification weight assigned to the scanned page;
- after determining two subsequent initial pages of two subsequent documents in the plurality of scanned pages, appending the scanned pages between the two subsequently determined initial pages to a first determined initial page;
- and
- creating an electronic file for each determined initial page and the scanned pages appended thereto.
11 more elements
A method of categorizing a scanned page without the use of separator pages, comprising:
- extracting content from the scanned page;
- assigning, based upon the extracted content and a layout of the scanned page, a classification weight to the scanned page;
- determining, based upon the assigned classification weight, if the scanned page is an initial page of a first document, wherein the determining comprises:
- assigning a classification weight to the scanned page based upon an analysis of the extracted content, wherein the classification weight reflects a discrete probability that the scanned page is one of the initial pages of the first document;
- and
- determining, based upon a comparison of the assigned classification weight to a sample set of classification weights, whether the scanned is an initial page of the first document;
- wherein the comparison comprises calculating a percentage probability of whether the scanned page is an initial page of the first document according to a formula:
- (1−((Wtype X−WAssigned)/Wtype X));
- wherein Wtype X is a classification weight of a page of a document of a particular type in a sample set of documents;
- and
- wherein Wassigned is the classification weight assigned to the scanned page;
- and
- if the scanned page determined to be an initial page of a first document, designating the scanned page as the initial page of the first document;
- appending at least one subsequently scanned page to the initial page of the first document after an initial page of a second document is determined;
- and
- creating an electronic file, the electronic file including the initial page of the first document and the at least one subsequently scanned page appended thereto.
13 more elements
A non-transitory computer readable storage medium storing one or more instructions that, when executed by a computer, cause the computer to perform a method of processing a plurality of page images to separate the plurality of page images into a plurality of documents without the use of separator pages, the method comprising:
- receiving the plurality of page images from a scanner;
- extracting content from each of the plurality of page images;
- for each of the plurality of page images, determining, based upon the extracted content, whether the respective page image is an initial page image of one of the plurality of documents wherein the determining comprises:
- assigning a classification weight to each of the plurality of page images based upon an analysis of the extracted content, wherein each classification weight reflects a discrete probability that the respective page image is one of the initial pages of one or more of the plurality of documents;
- and
- determining, based upon a comparison of the assigned classification weight to a sample set of classification weights, whether the page image is an initial page of a document in the plurality of documents;
- wherein the comparison comprises calculating a percentage probability of whether the page is an initial page of one of the plurality of documents according to a formula:
- (1−((Wtype X−WAssigned)/Wtype X));
- wherein Wtype X is a classification weight of a page of a document of a particular type in a sample set of documents;
- and
- wherein Wassigned is the classification weight assigned to the page image;
- and
- appending each of the plurality of page images located between a first determined initial page image of the one of the plurality of documents and a subsequent, second determined initial page image of another of the plurality of documents to the first determined initial page image.
10 more elements
Dependencies are inferred from the claim text.