 |
 |
|
|
|
|
|
|
|
|
|
 |
|
 |
 |
|
|
|
|
|
|
|
|
|
Learn more about A2iA's automated intelligent document classification, data extraction & search tools for e-discovery. |
|
|
|
|
|
|
|
|
|
 |
 |
 |
|
|
|
 |
|
|
|
 |
|
|
A2iA products integrate the following proprietary technologies, which are developed and deployed within the company's R&D department |
|
Intelligent Word Recognition (IWR) |
|
Recognition of unconstrained handwritten words. Instead of recognizing words character- by-character, IWR software recognizes entire handwritten words. IWR technology matches handwritten or printed words to a user-defined dictionary, significantly reducing character errors encountered in typical character-based recognition engines. |
|
A confidence value is calculated for every word within a specific vocabulary list in order to determine the word(s) most likely to be inscribed in the image. The vocabulary can be supplied through A2iA's products or defined by the engine's user so as to adapt it to the specific context of the application processed. For each word analyzed, the system considers various shape and letter groupings in order to calculate the confidence value associated with the word in question. |
|
Intelligent Character Recognition (ICR) |
|
Recognition of handwritten or machine-printed, omnifont characters. ICR software recognize isolated characters by distinguishing a wide variety of shapes and sizes within each character classifier. Rather than using reference character images, these systems draw on extracted features such as curves, loops and lines. The characters must first be extracted from the data fields. Several character segmentation methods are used by A2iA's recognition engines. ICR techniques are employed when the characters are relatively well formed and distinctly separated. |
|
Document management |
|
To locate data fields within the pages of digitized documents and to distinguish handwritten information from the document's template, A2iA has developed specialized document management modules for check processing and address recognition, as well as modules for structured documents and forms processing. These modules can also be configured for fixed formats. These modules rely on the detection of key printed words and geometric forms such as lines and boxes, and on the alignment of the structured document against the document template. |
|
|
|
|
|
|
|
|
|
 |
|
|