Multi-page document recognition in document capture

Abstract

Techniques to capture document data are disclosed. It is determined that a sequence of pages in a stream of document page images comprise a single multi-page document. Data is extracted from two or more different pages included in the sequence. The data extracted from two or more different pages included in the sequence of pages is used to populate a data entry form associated with the multi-page document.

What is claimed is: 
     
         1 . A method of capturing document data, comprising:
 obtaining a multi-page document;   extracting data from multiple pages of the multi-page document;   identifying two or more values located on at least two different pages of the multi-page document, wherein at least a first one of the two or more values is dependent on at least a second one of the two or more values; and   validating the at least the first one of the two or more values and the at least the second one of the two or more values according to one or more validation rules.   
     
     
         2 . The method of  claim 1 , wherein the validating comprises: accessing a set of validation rules in a library of validation rules; sequentially applying the set of validation rules to the at least the first one of the two or more values and the at least the second one of the two or more values; and marking the at least the first one of the two or more values and the at least the second one of the two or more values as having an error if one of the validation rules fails. 
     
     
         3 . The method of  claim 2 , further comprising receiving a document type definition corresponding to the multi-page document, wherein the document type definition identifies the set of validation rules to be applied to the at least the first one of the two or more values and the at least the second one of the two or more values. 
     
     
         4 . The method of  claim 3 , wherein the document type definition includes a mapping to document type fields to be used to apply each rule. 
     
     
         5 . The method of  claim 4 , further comprising identifying a document type of the multi-page document, and identifying the document type definition based on the identified document type of the multi-page document. 
     
     
         6 . The method of  claim 5 , wherein the document type contains one or more scalar fields and one or more tables of array fields. 
     
     
         7 . The method of  claim 6 , further comprising extracting values from each page into per-page scalar and array fields by name. 
     
     
         8 . The method of  claim 7 , wherein for each extracted value, a corresponding location on the page from which the value was extracted is saved. 
     
     
         9 . The method of  claim 6 , further comprising combining data extracted from the respective pages into a form associated with the document type, wherein combining data extracted from the respective pages into a form associated with the document type includes forming an array that spans multiple pages concatenating a first set of rows of values extracted from a first page with a second set of rows of values extracted from a second page to create a combined set of rows to be included in the document type. 
     
     
         10 . The method of  claim 1 , wherein the validating comprises providing one or more of the at least the first one of the two or more values and the at least the second one of the two or more values to a user for manual validation. 
     
     
         11 . The method of  claim 10 , further comprising presenting an interface to the user, wherein the interface displays to the user a plurality of fields which are marked as having errors and enables the user to iterate through the plurality of fields, wherein the plurality of fields displayed to the user include only dependent fields that require validation. 
     
     
         12 . The method of  claim 11 , further comprising identifying a document type of the multi-page document, and creating an instance of a selected one of a plurality of type-specific data entry forms in a forms library based at least in part on the identified document type of the multi-page document, wherein the plurality of fields displayed to the user are fields contained in the created instance. 
     
     
         13 . The method of  claim 12 , wherein the two or more values presented to the user for manual validation are identified by determining whether the first one of the two or more values matches the second one of the two or more values in the multi-page document. 
     
     
         14 . The method of  claim 11 , wherein the interface is configured to repetitively iterate through each of the plurality of fields until either the user enters data that clears the validation error associated with the field. 
     
     
         15 . The method of  claim 11 , further comprising populating, based at least in part on the data extracted from the multi-page document, a plurality of fields of the instance of the selected data entry form including the plurality of fields displayed to the user. 
     
     
         16 . The method of  claim 11 , wherein as each form field is displayed, a corresponding snippet or other partial image from a page from which a current data value associated with the form field was extracted is displayed adjacent to the field. 
     
     
         17 . The method of  claim 1 , wherein the first and second ones of the two or more values are contained in a table or array of the multi-page document. 
     
     
         18 . The method of  claim 1 , The method further comprising determining that a sequence of pages comprise the multi-page document, wherein determining that a sequence of pages in a stream of document page images comprise the single multi-page document includes processing each page individually to determine a corresponding page type; and processing the stream of page types to identify a sequence associated with a multi-page document type. 
     
     
         19 . A document capture system, comprising:
 a communication or other interface configured to receive a multi-page document; and one or more processors coupled to the interface and configured to:
 obtain a multi-page document; 
 extract data from multiple pages of the multi-page document; 
 identify two or more values located on at least two different pages of the multi-page document, wherein at least a first one of the two or more values is dependent on at least a second one of the two or more values; and 
 validate the at least the first one of the two or more values and the at least the second one of the two or more values according to one or more validation rules. 
   
     
     
         20 . A computer program product to capture document data, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:
 obtaining a multi-page document;   extracting data from multiple pages of the multi-page document;   identifying two or more values located on at least two different pages of the multi-page document, wherein at least a first one of the two or more values is dependent on at least a second one of the two or more values; and   validating the at least the first one of the two or more values and the at least the second one of the two or more values according to one or more validation rules.