OCR software developed for the production marketplace, PrimeOCR is a Windows based software OCR engine that uses "Voting" technology to reduce conventional OCR software error rates by 65-80%! We license the best retail OCR engines so you don't have to. We pass your image through each engine and use Voting technology along with Artificial Intelligent algorithms to achieve 65-80% better accuracy than conventional OCR software!
The best conventional OCR software products achieve about 98% average accuracy recognizing text on typical quality document images. While 98% accuracy sounds impressive, that still leaves 40 errors on a typical 2000 character text page! These errors must be corrected manually, which is expensive. As a result, many imaging installations that want to use OCR software cannot afford it.
Installations that use PrimeOCR can cut errors per page by 65-80% (down to 8 errors per average 2000 character page) thus reducing the need for error correction by 65-80%. This translates into not only 65-80% less manual error correction but also 65-80% fewer error correction workstations.
Although you pay more for PrimeOCR than a conventional OCR engine, the total system cost is lower because PrimeOCR dramatically reduces the need for error correction. More importantly, you save 65% of the annual manual error correction cost, which in OCR intensive applications is often 50-70% of the total imaging system costs.
And PrimeOCR saves time even if you don’t perform any OCR cleanup and instead use a "Fuzzy" search engine to make up for OCR software errors.
Can a Fuzzy Search Engine make up for OCR mistakes?
Fuzzy Search engines try to account for OCR errors by searching a database for documents containing words that are similar to the key word(s) you supply. THE PROBLEM - too many documents are returned for your review. The document you are looking for may be in the list, but reviewing 10's, even 100's of documents to find it takes time. The more accurate your OCR, the more exact matches a fuzzy search can find and the fewer documents you need to review. The more accurate your OCR, the more likely the data you are looking for will appear at the top of a fuzzy search "hit" list. This saves time by allowing you to find your data faster.
This is especially true for databases with over 1,000 documents, where the number of false hits can be so large that reviewing all search results is too time consuming.
Only 40-60% of errors generated by standard OCR software products are "flagged" for correction. Since manual editing typically corrects only flagged errors, as many as 60% of the errors produced do not get corrected. They end up going into the database or application undetected. By reducing the total number of OCR generated errors, and generating more accurate suspicious character “flags”, PrimeOCR also reduces the total number of errors that make it into your database or final application by 75%.
PrimeOCR - High Fault Tolerance Through Automatic Engine Recovery
Ever submit images for overnight OCR processing, only to find in the morning that the OCR product failed sometime during the night due to 1 bad image? Your remaining images have not been OCR'd and the same thing could happen all over again the next night!
Enter PrimeOCR''s Automatic Engine Recovery.
A very poor quality image can cause a conventional OCR product to "crash". Typically, the only way to reinitialize the OCR software is to reboot the PC. Not only is this time-consuming, it requires that someone constantly monitor the OCR process. To solve this problem, PrimeOCR has a new feature called Automatic Engine Recovery. PrimeOCR will sense when an engine fails and automatically reinitialize it for the next image. No downtime, no manual intervention and 2/3rds fewer errors than conventional OCR - Only available through PrimeOCR!