Computer Vision vs. OCR vs. CMR

Using machine learning to make computers see

Add bookmark

IAN Editorial Team
03/02/2022

What is Computer Vision?

Computer vision is a field of artificial intelligence (AI) focused on training computers to interpret and understand the visual world. The goal is to enable computers and systems to derive meaningful information from digital images.

There are numerous ways computer vision can be configured. Oftentimes unstructured data is captured via camera or sensor then routed into a data ingestion engine where it is processed and classified. Two of the most common data ingestion engines are optical character recognition (OCR) and cognitive machine reading (CMR).

Unstructured Data Processing Industry Survey

Tell us how your organization transforms unstructured data into machine-readable insights by participating in our latest industry survey.

https://www.surveymonkey.com/r/WC3YP33

What is Optical Character Recognition (OCR)

Optical character recognition (OCR) is a technology that automates data extraction from different types of documents, such as scanned paper documents, PDF files or images captured by a digital camera.

OCR was designed to recognize machine printed, fixed field text while its offshoot, intelligent character recognition (ICR), is used to decipher handwritten text.

Some of the benefits of OCR and ICR include:

Elimination of manual data entry
The ability to process more data faster and with fewer resources
Increased accuracy
Reduced physical storage space
Enables the automation of tasks

One common application of ICR and OCR is check processing. If you’ve ever deposited a check using an ATM or mobile app, chances are a combination of ICR and OCR solutions were used to “read” and process your check.

For decades, organizations have relied on OCR to automate document processing. Though the technology has certainly proven its value over the years, it is not without its limitations. Most OCR and ICR solutions can only process fixed field text and every time a new type of document is processed, a template must be uploaded. In addition, it cannot extract non-text data such as images or charts/graphs.

As such, organizations are increasingly embracing cognitive machine reading (CMR), the next generation of intelligent document processing solutions.

Cognitive Machine Reading (CMR)

While OCR/ICR is used to extract text data, cognitive machine reading (CMR) can process unstructured data from a wide range of sources including checkboxes, tables, handwriting, cursive, and images. In other words, it uses Artificial intelligence (AI) and machine learning (ML) techniques to recognize, ingest and classify multiple data formats within one platform.

Compared to OCR and ICR, CMR is significantly faster. While OCR uses sequential and linear character matching, CMR leverages just-in-time pattern recognition techniques. In addition to speeding up data extraction, this technology also ensures you only capture the data you need, nothing else.

Another major benefit of CMR is that it eliminates the need for manual intervention and enables straight-through processing. It also replaces supervised learning with automated learning, a.k. self-learning. Using advanced machine learning (ML) techniques, CMRs can learn and self-improve on their own, dramatically reducing the oversight needed to run them.