A major British retailer has implemented an OCR solution to automate its stock inventory system. Dutch company Prime Vision has supplied its OCRlite software to check boxes of stock as they are delivered to the retailer's distribution warehouse.
The majority of information is contained within a barcode but suppliers manually write the number of items in each box on the side of the container. OCRlite was used to identify the handwritten numbers on the boxes to automate the stock-taking process.
A large amount of research goes into providing scanners that read handwriting effectively, a process known as Intelligent Character Recognition (ICR), and, according to Mark Ryder, international sales manager at Prime Vision, few companies have the resources and expertise in developing software for identifying handwritten text. Some machine vision software packages include recognition tools for handwriting, such as the Common Vision Blox (CVB) Manto from Stemmer Imaging.
OCR is used in many industries from mail sorting to pharmaceutical and while accuracy rates for reading printed text are high, using software to read handwriting is more difficult. 'The challenge with reading handwriting is in separating the characters,' explained Ryder. In addition, a computer doesn't understand context as a human would. 'The software can be supplied with a context, for instance, if there are always a set number of characters, such as a postcode, this can be used to aid character recognition,' he said.
According to Ryder, the number of manufacturers providing cameras for OCR solutions has grown and a greater variety of camera systems are available in the marketplace. The camera systems must provide high-resolution images, to identify small font sizes, and operate at high speed, to capture images of items moving rapidly on conveyor belts.
Ryder said that users are demanding systems that can identify a broader range of items, including logos, graphics, handwriting, and machine printed text, and that are more configurable to specific requirements.
