- OCR tools scan documents into bi-level images, treating light areas as background and dark areas as characters, or glyphs.
- OCR identifies glyphs with two algorithms: pattern recognition and feature detection.
- Feature detection uses rules per character, such as angled lines, crossed lines and curves.
- OCR results are cross-referenced against an internal dictionary for accuracy.
- Banks use OCR to verify deposit checks and loan documents; healthcare uses it to digitize patient records.
How do OCR Tools Work?
OCR tools scan physical documents and convert them into black-and-white versions called bi-level document images. These scanned images are analyzed for light and dark areas. The light areas are categorized as background and the dark areas are categorized as image characters or glyphs. Glyphs are identified using two algorithms: pattern recognition and feature detection.
Pattern recognition isolates the character image, or glyph to compare examples of other stored glyphs in different formats and fonts.
Feature detection is a set of rules created for each letter or number to identify characters. Features include angled lines, crossed lines, or curves present in the character. These features are used to find the best match among stored glyphs.
After pattern recognition and feature detection, the results are cross-referenced for accuracy using an internal dictionary.
OCR Use Cases
Financial services
The financial services industry leverages OCR to process and verify paperwork for loan documents, deposit checks, and process other financial transactions. OCR allows real-time verification of deposits via check. The OCR system creates the data that machine learning algorithms analyze to detect suspicious transactions.
Healthcare
The healthcare industry uses OCR to grant patients and doctors with digital access to health records, including X-rays, treatment, test results, and insurance payments. OCR enables these documents to be scanned, processed, and stored across healthcare databases. This reduces manual labor and streamlines workflow across hospitals while keeping records up to date.
Benefits of OCR
OCR automates the documentation process and improves accuracy. Time and resources are saved by removing human error from the manual documentation process.
Other benefits include:
Higher productivity
Increased accuracy
Superior data security
Text-searchable documents
Improved customer service
Easily editable documents
Optical Character Recognition in H2O AI
H2O Document AI combines OCR with natural language processing to build models that extract text and fields from scanned documents and images.