Prime Vision has launched the latest generation of its Find-It region-of-interest finder, designed for mail sorting. The software tool is available as a modular component that integrates into the company’s ParcelMatch and MailMatch OCR platforms. Neural network technology allows it to be trained for specific label or parcel characteristics and for multiple strategies; it can be programmed to spot barcodes, logos or keywords or as a label identification tool.
If address details are not found automatically, processing time and costs rise exponentially. The rejected mail item becomes subject to manual processing, an outcome that posts are striving to minimise in order to contain costs, particularly where high volumes are involved.
For parcel applications in particular, automatically finding the region-of-interest is a complex task. The total image size may be up to 1m square and the parcel is multi-faceted. It may also carry a number of labels and other conflicting information to prevent the OCR doing its job.
Prime Vision’s Find-It has been designed to select the correct region-of-interest for the OCR engine to process. It also works across multiple surfaces – on a box, for example, it can spot which of its six sides carries the address data.
