Drawing OCR Services: make your drawings searchable!
A different kind of drawing conversion: low cost, bulk service to create searchable PDF files
The life of a drawing does not stop at the CAD program: it will be used over the file of the representative product over and over. The fact is that even though there are high quality vector drawings available for the same part, building or product, usually the paper-equivalent drawing will be used in business processes.
This is due to many reasons, but the most common will be the proprietary nature of the content, the manufacturing process, logistics and version control. The result is that we are ending up owning lots of paper drawings or scanned/cad generated equivalents.
The best practices in document management call for searchable content based on metadata (Drawing No.. Title, Revision, Used on, etc...) and content ().
Unfortunately, general-purpose programs are not designed to address the specific challenges presented by text entities on drawings. Some of these are:
- Drawing fonts: many time these are hand-written
and specific to an industry segment (DOD drawings, architectural,
facilities, etc...) that are not handled by the general-purpose
routines of a standard package.
- Text Orientation: There are at least horizontal
and vertical text entities of interest on many drawings (labels, part
numbers circuit names etc...).
- Text-like drawing entities: many drawings contain symbols that are similar to characters, greatly confusing the machine processes.
- Large drawing sizes: usually there are limitations in drawing canvas sizes that a standard engine or program can handle.
How are we overcoming all these hurdles here at ELAN GMK?
Proprietary technologies, using special routines (and some common sense!) make it all possible to address all of the above obstacles. Some of these technologies are:
- Page segmentation before to select only text
entities form the drawing. This is a crucial technology developed by
ELAN GMK. Only text segments will be submitted to a special text
conversion engine, thus eliminating size limitations
- Special engine: proprietary, custom-developed
engine to address this market segment. Includes and ICR
- Multiple-pass : the same drawing will be passed through several orientations and optimized routines select the best results for each text entity.
- Server-based implementation: The advanced image
processing requires a lot of resources, some drawings might need longer
time to process (average 30 seconds per drawing) that is not practical
for desktop usage. Because of this, all the conversion processes are
implemented as Windows Services that can be run automatically (server
mode) on several PCs. The throughput is only limited by the number of
processor machines available, making the solution highly
There are some limitations to this process. All the operations are automatic, and the results are left uncorrected. We have to mote that this is a low cost service designed to give you immediate results for large volumes of drawings.
Please take a look at our questionnaire by clicking on the link below and see if we can help you in any way.