Dynamsoft unveils label recognition technology

Dynamsoft is offering a software development kit for application developers that enable users to scan and extract content digitally from physical labels. This is ideal for expanding data capture capabilities in automotive, government, retail, warehousing, and other industries.

This technology can aid with things like ensuring pricing accuracy or verifying physical assets, providing for more efficient workflows and improved productivity, according to the vendor.

“One example where Dynamsoft Label Recognition is ideal is in a grocery store, where prominent price labels are often used to call out a sale,” said Amy Gu, president, Dynamsoft. “Such labels do not always match the price on the barcode, resulting in price checks during checkouts, which are frustrating customer scenarios all around. This also costs consumers and stores millions in losses annually. Our technology makes it simple for an employee to move down an isle to scan multiple labels and barcodes and instantly call out any mismatches.”

Dynamsoft Label Recognition SDK is based on optical character recognition (OCR) technology. OCR is widely used for extracting text from scanned paper documents so it can be digitally saved and edited.

However, for scan accuracy and speed efficiencies in specific industries, Dynamsoft has optimized its OCR implementation for use with short and sparse content, such as found on labels. So, Dynamsoft Label Recognition SDK provides developers with specific optimization options.

Developers can provide users with an option to run zonal OCR in a text area. With this, a user can specify a relative position for a reference region.

A regular expression can also be set to inform Dynamsoft Label Recognition of specific content structure. For example, a VIN code has a fairly fixed area in terms of size, and each digit has certain values that are to be used. While it is likely in most cases the SDK can scan such a VIN without issue, some environments can pose challenges.

Also, for images of poor quality, sophisticated algorithms are applied to improve contrast, remove noise, and so on. The characteristics of content can also be used to improve image quality. For example, the curve of text on a circular object can be used to correct distorted images.

Finally, Dynamsoft Label Recognition supports reading text from video streams. For a particular frame, the SDK extracts the content using OCR and cross references the result with neighbor frames for the best accuracy. If the SDK decides that the text result is partial, it intelligently stitches the results together.

For more information about this news, visit www.dynamsoft.com.