Prime Recognition Product Architecture
Out of the Box Results
Using standard, off the shelf PCs, Prime Recognition products are easy to install. Users are up and running within minutes. Although rich in features and built to process images reliably, the PrimeOCR Job Server has been designed to be very simple. Default settings are defined to simplify the installation.
The absence of a processing script language, dedicated hardware, complicated compression schemes, proprietary binary formats, or third party databases makes the integration of Prime Recognition products into any imaging workflow seamless.
Typical Configurations
Prime Recognition products can run on stand alone PCs or on PCs connected to any network. Since each application is file based, applications can run anywhere on a LAN/WAN network as long as the application can access the directory where the images are stored.
Most customers license PrimeOCR as a stand alone PC server. They may license just one copy or several copies to meet their throughput demands. PrimeOCR is licensed on a per PC basis.
Customers that need very high accuracy license PrimeView to manually zone text areas of an image prior to OCR and license PrimeVerify to verify low confidence characters after OCR. PrimeView generates temple files that are used by PrimeOCR and PrimeOCR generates .PRO output that is used by PrimeVerify for character review.
Any number of users create jobs from the network with PrimeView or with a user application or the PrimeOCR Job Server Wizard which can be used as a stand alone application. Images are loaded from a network directory, possibly a NFS mounted drive from a UNIX server, or an Windowsserver. Jobs are created and saved to a user specified directory on the network or on the Job Server PC. The PrimeOCR Job Server, running Windows, queues the job files for processing. The results from OCR processing can then be saved to another, or the same user specified directory on the network, where post OCR clean up can take place.
Scalable and Reliable
Higher throughput can be achieved by simply adding additional PC's to the network. Depending on the imaging application, an additional PrimeOCR Job Server, and/or several additional PrimeVerify workstations may be required.
Scaling the system for increased throughput is as easy as hooking up standard PCs to the network, installing the software, and pointing the applications to the same user defined directories being processed. No programming is required, just a simple point and click to select the input job directories, and output directories and additional high accuracy OCR processing resources are available to the imaging system.
Further OCR processing power can be harnessed by using standard high speed, multi-processor systems to run the PrimeOCR Job Server. Prime Recognition products do not rely on dedicated computing hardware or network protocols.
Because the PrimeOCR Job Server is file based, the increased complexity of relying on third party applications such as a database is minimized and increases processing reliability. Each image is read into the PrimeOCR engine one image at a time. This reduces memory requirements and provides reliable processing for thousands of images through the server.
