Image Cleaning Solution for Legal Industry
Cooperation between ATAPY Software and EasyData B.V. continues to bring new advantages to Benelux customers
‘Raad voor Rechtsbijstand’ (www.rvr.org) is a Dutch agency that provides legal advice and public prosecutors to those people who are unable to hire a lawyer of their own. The agency maintains a large archive of various legal records. While attempting to convert the archive to a digital format, the agency faced with an unexpected problem.
Many documents were printed on intensively colored paper. The same background that made the paper documents look nice and distinct, on black-and-white scans turned up as a heavy jitter, inflating the file sizes, making them difficult to read and impossible to OCR.
The agency sought professional advice from EasyData B. V. Having analyzed the issue, EasyData concluded that no ready-made solution, such as despeckling facilities of the modern OCR packages, was able to produce acceptable results, as the jitter was significantly heavier than what those tools were capable of removing. EasyData outsourced the problem for a more thorough research to ATAPY Software. ATAPY’s engineers designed and implemented a custom algorithm that approached the task more intelligently, taking into account not only the linear characteristics of each dot cluster but also its context (characteristics of the neighboring clusters). The result was a tool which produced nearly-clean images that took up to 10 times less disk space, and with only a very insignificant information loss:
The original image (OCR rate after built-in despeckling = 1.9%):
The same image after ATAPY’s cleaning algorithm has been applied (OCR rate = 98.9%):
About Raad voor Rechtsbijstand:
The Raad voor Rechtsbijstand (Legal Aid Board) is a Dutch agency that provides legal advice and public defenders to people who are unable to hire a lawyer on their own. For more information visit www.rvr.org.